Dec 08, 2025 Technology and Security Committee

ARE

[00:00:01]

[1. Call General Session to Order]

UM, HELLO, I'M JOHN SWENSON.

I'M THE COMMITTEE CHAIR.

UM, AND I CALL THIS MEETING TO ORDER.

THIS MEETING IS BEING WEBCAST LIVE TO THE PUBLIC ON ER O'S WEBSITE, AND I HAVE CONFIRMED THAT THERE IS A QUORUM IN PRESENT.

THERE'S NO QUORUM OF THE PUBLICITY COMMISSION.

SO, UH, THEY ARE HERE, HERE AS OBSERVERS, UM, WHICH IS GREAT, UM, FOR MOVING ON TO, HELLO? CAN YOU HEAR ME? GOOD MORNING, DR.

SHAY.

HEAR YOU.

CAN YOU HEAR US? MAYBE YOU HAVE TO ALLOW YOUR MIC TO WORK.

MY MIC IS WORKING.

SO MAYBE YOUR SPEAKER THAT'S NOT SO OUR OUR PEOPLE ARE LOOKING AT IT RIGHT NOW.

WE'LL JUST, UH, LOOKS LIKE YOU CAN HEAR ME, BUT, UH, WE CAN HEAR YOU.

I'M FORWARD TO HEARING YOU, BUT YOU CAN'T HEAR US.

CANNOT HEAR YOU YET.

THAT'S A PROBLEM FROM MY END.

IT SEEMS LIKE ALL OF YOU ARE MUTED.

YES, I DO SEE AT THE BOTTOM OF THE SCREEN.

AND THE MUTE BUTTON IS YES, I SEE THAT TOO.

HEY, DANA, DO YOU SEE THE BOTTOM OF THE SCREEN IN FRONT OF YOU? THAT MUTE BUTTON? THE MUTE BUTTON IS ON RIGHT THERE.

YEAH.

DR.

SHAA, CAN YOU HEAR US NOW? NOW I CAN HEAR YOU CLEAR AND WELL.

WONDERFUL, WONDERFUL.

SMALL MATTER OF PUSHING THE RIGHT BUTTON.

UM, LET'S KEEP GOING.

UH, WE HAVE, UH, THE, THE FIRST ORDER OF BUSINESS, UH, IS THE, UM, ANTITRUST ADMONITION SECURITY MAP, WHICH SHOULD BE IN FRONT OF YOU.

UH, I'LL LEAVE THAT FOR YOUR EDIFICATION.

UM, WE HAD THE, THE SECOND ORDER OF BUSINESS.

ITEM TWO IS NOTICE OF PUBLIC COMMENT, IF ANY.

TODAY'S MEETING AGENDA WAS POSTED PUBLICLY ON DECEMBER 1ST AND PROVIDED INSTRUCTIONS FOR THE PUBLIC TO COMMENT IN PERSON.

TO DATE, NO ONE HAS EXPRESSED AN INTEREST.

IS THAT STILL CORRECT, BRENT? THAT IS CORRECT.

CHAIR.

THANK YOU VERY MUCH.

THEN LET

[3. September 22, 2025, General Session Meeting Minutes]

UM, DOES ANYONE HAVE ANY COMMENTS OR WISH TO MAKE A MOTION? CHAIRMAN, I WISH TO MAKE A MOTION.

THANK YOU VERY MUCH, SECONDER.

I'LL SECOND IT.

THANK YOU, LINDA.

ALL IN FAVOR? AYE.

AYE.

ANY OPPOSED? MOTIONS CARRIED.

UM, IT BRINGS

[4. Guest Presentation: A Roadmap Toward Agentic Intelligence in Power Systems]

UM, I'M DELIGHTED TO, UH, INTRODUCE, UH, DR.

LECHE, UH, WHO'S A PROFESSOR AT HARVARD, UH, AND WHO'S JOINING US REMOTELY, ALONG WITH HIS GRADUATE STUDENT, UH, IAN ZANG.

UM, DR.

SHEA IS GORDON MCKAY, PROFESSOR OF ELECTRICAL ENGINEERING AT HARVARD UNIVERSITY.

UH, HE SERVES ON THE BOARD OF MANAGERS OF THE PJM INTERCONNECTION, OUR, OUR FELLOW ISO, UP IN THE PACIFIC NORTH, OR THE ATLANTIC COAST, I SHOULD SAY.

UM, HE RECEIVED HIS BACHELOR IN ELECTRICAL ENGINEERING FROM TINGO UNIVERSITY IN 2004, MASTER'S IN ELECTRICAL SCIENCES FROM HARVARD IN 2005, AND A PHD FROM CARNEGIE MELLON IN 2009.

HIS INDUSTRY EXPERIENCE INCLUDES WORK WITH ISO NEW ENGLAND AND EDISON MISSION ENERGY TRADING AND ENERGY MARKETING AND TRADING.

UM, UH, CHAN ZANG IS A PHD STUDENT IN ELECTRICAL ENGINEERING AT HARVARD UNIVERSITY.

HE'S A CO-FOUNDER OF THE POWER AGENT OPEN SOURCE COMMUNITY, WHICH IS DEDICATED TO ACCELERATING THE DEVELOPMENT OF AG AGENTIC AI, AND POWER SYSTEMS. WE THANK OUR GUESTS FOR JOINING US TODAY AND NOTE THAT COMMITTEE MEMBERS WILL HAVE A CHANCE, UH, OPPORTUNITY TO ASK ANY RISK MANAGEMENT RELATED QUESTIONS DURING THE EXECUTIVE SESSION THAT, THAT WILL FOLLOW THIS.

SO, WITH THAT, PLEASE PROCEED.

DR.

SHANE.

THANK YOU.

UH, IT'S A GREAT PLEASURE AND HONOR FOR ME TO BE HERE, HERE WITH YOU.

UH, I MUST FIRST BY SAY, UH, UH, I WOULD VERY MUCH LOOKED FORWARD TO, TO MEETING YOU IN PERSON.

UH, BUT, UH, JUST AS WE ARE HAVING A BOARD MEETING HERE NOW AT ERCOT, UH, WE HAD A BOARD MEETING FOR HARVARD, UH, LAST NIGHT, WHICH I COULD NOT EXCUSE MYSELF OF.

SO APOLOGIES FOR NOT BEING THERE, UH, IN PERSON.

BUT I MUST SAY, UH, LONG BELATED HOWDY, UH, BECAUSE I, I SPENT 15, UH, YEARS OF MY LIFE IN COLLEGE STATION AND

[00:05:01]

SO ERCOT HAS BEEN ALWAYS, UH, A POINT OF A BEACON OF INNOVATION AND, AND, AND, UH, UH, TACKLING REAL WORLD CHALLENGES FOR US.

SO, UH, APPRECIATE ALL YOU DO AND, UH, THANK YOU FOR YOUR SERVICE.

UH, TODAY, I'M VERY PLEASED TO, TO SHARE WITH YOU, UH, SOME OF THE ONGOING WORK AT THE INTERFACE OF, OF AI AND POWER SYSTEMS, UH, AND, AND IN PARTICULAR, UH, ALONG THE LINE OF THE MOST, UH, RECENT, UH, PROGRESS ON, UH, AGE AGENTIC AI, UH, WHICH IS SORT OF A ONE, ONE OF THE MOST, UH, CHALLENGING AND AND EXCITING OPPORTUNITIES, UH, FOR, FOR, FOR THE POWER INDUSTRY TO, TO GRAPPLE WITH.

UM, GOING TO MOVE ON TO THE NEXT PAGE.

SO HERE AT HARVARD, WE, WE HAVE BEEN THINKING LONG AND HARD ABOUT THIS INTERPLAY, UH, BETWEEN AI AND, AND AT POWER SYSTEMS, AND PERHAPS MORE GENERALLY SPEAKING, COMPUTING AND, AND POWER INDUSTRY.

UH, WHAT WE FOUND, YOU KNOW, VERY INTERESTING IS THAT, YOU KNOW, THESE TWO AREA WHICH SORT OF USED TO EVOLVE ON ITS OWN AND IN SORT OF A SOMEWHAT, UH, PARALLEL PATHS ARE BECOMING MORE AND MORE, UH, INTERACTIVE AND MORE AND MORE INTEGRATED.

AND BY THAT I MEAN, UH, WE AS POWER ENGINEERS, UH, NEED TO KNOW A LOT MORE ABOUT HOW TO IZE THE, UH, GREAT POTENTIAL THAT AI HAS TO OFFER.

AND, AND IN THE MEANTIME, WE AS POWER ENGINEER HAVE TO THINK A LOT ABOUT, AND ERCOT IS A FRONTLINE, UH, UH, YOU KNOW, UH, LEADER IN, IN THIS ISSUE OF HOW, HOW, HOW ARE WE GOING TO, YOU KNOW, MAKE SURE THAT, UH, ALL THIS AI AND COMPUTE, UH, COULD VERY WELL BE INTEGRATED THE POWER GRID SO THAT, UH, WE CAN KEEP THE LIGHTS ON, UH, WHILE IN THE MEANTIME SUPPORT THE GROWTH OF, OF ARTIFICIAL INTELLIGENCE.

AND, UH, THE WAY WE WOULD LIKE TO CONCEPTUALIZE THIS INTERPLAY IS PERHAPS THROUGH A, A, A LENS OF, OF TIMESCALES.

SO BOTH IN TERMS OF, UH, SORT OF A REAL TIME OPERATIONAL TIMESCALE, AS WELL AS ON THE SIGHTING AND PLANNING TIMESCALE, UH, WE CAN SEE A LOT OF, UH, UH, FRUITFUL INTERACTIONS, UH, BETWEEN POWER GRID AND, AND AI OR COMPUTING.

AND ONE OF THE UNIQUE ANGLE THAT BRINGS THESE TWO INDUSTRY CLOSER TO EACH OTHER IS THE EMERGENCE OF FOUNDATION MODELS.

AND FOR THOSE OF YOU WHO MAY NOT, YOU KNOW, BE WELL VERSED IN THIS NOTION OF FOUNDATION MODELS, LET ME JUST SAY THAT FOUNDATION MODEL, IT IS EXACTLY THE, THE KINDS OF MODEL THAT POWERS CHAT, GPT AND, AND GEMINI AND, AND, AND, AND SO ON, RIGHT? THIS IS A, UH, LARGE LANGUAGE, TRANSFORMER BASED MODEL THAT POWERS MORE THAN ONE TASK, VERY, VERY, UH, EFFECTIVELY, RIGHT? AND, UH, ON THE SHORTER OPERATIONAL TIMESCALE, ONE OF THE VERY CHALLENGING ISSUE THAT I THINK ALL OF US ARE, ARE GRAPPLING WITH IS THAT WHAT EXACTLY IS THE, A PHYSICAL, YOU KNOW, POWER PROFILE THAT THE AI DATA CENTER, A LIKELY GOING TO PRESENT THEMSELVES TO THE POWER GRID? YOU KNOW, IN OUR MIND, IT MAY FEEL LIKE A FLAT LINE, BUT IF YOU LOOK UNDER THE HOOD IN A LOT CLOSER, UH, YOU CAN SEE THAT, UH, THE POWER PROFILE OF A TYPICAL AI TRAINING DATA CENTER IS FAR FROM BEING FLAT, RIGHT ON THE TOP RIGHT, YOU SEE A STATE OF THE ART, UH, META AI TRAINING POWER PROFILE.

WHAT DOES IT LOOK LIKE? IT REMIND ME A BIT OF A, A WIND AND SOLAR FLUCTUATIONS, RIGHT? ALTHOUGH IT'S, UH, PERHAPS IN A, IN A SLIGHTLY DIFFERENT TIMESCALE HERE, RIGHT? SO, SO THERE'S A LOT OF UNKNOWN THERE ON CHARACTERIZING THE INTRINSIC PHYSICAL POWER DYNAMICS DURING THE AI AND INFERENCE, UH, PERIODS AND AGGREGATE THEM ALL TOGETHER AT, AT THE WHOLESALE LEVEL.

ONE OF THE BIG UNKNOWN QUESTION, WHICH I THINK ERCOT IS FRONT AND CENTER ON, IS HOW ARE THESE LARGE LOADS GOING TO RESPOND TO, TO MARKET SIGNALS, RIGHT? ARE THEY GOING TO BE JUST AS FLEXIBLE AS WE WOULD HOPE FOR, OR ARE THEY GOING TO BE TOTALLY INFLEXIBLE? AND THE ANSWER IS PROBABLY SOMEWHERE IN BETWEEN.

UH, WE DON'T HAVE A CLEAR ANSWER TO THAT.

THAT'S PRECISELY, WE NEED

[00:10:01]

BUT, YOU KNOW, AS YOU CAN SEE, ONE OF THE, UH, EXAMPLE WE HAVE SHOWN HERE IS ACTUALLY A REAL DATA WE OBTAINED FROM, UH, UH, ABOUT 20 MONTHS AGO, JANUARY, UH, 2024, UH, WINTER STORM HEADER, UH, RIGHT BEFORE THE MARTIN LUTHER KINGS DAY, UM, WHEN, WHEN THERE WAS A, A, A LARGE A, A WEATHER INDUCED, UH, PEAK, WHICH LED ONTO A WHOLESALE MARKET PRICE SPIKE THAT YOU SEE HERE ON THIS, UH, UH, BLUISH CURVE.

AND THEN THE AREA THAT IS SHADED WITH GLIDE GREEN IS, UH, THE BEHAVIOR OF THIS LARGE LOWS BACK THEN, JANUARY, 2044, IT WAS LARGELY CRYPTO MINING.

UH, SO, SO YOU CAN SEE THEY ARE SOMEWHAT RESPONDING, BUT NOT ENTIRELY.

SO A QUESTION IS REALLY HOW DO WE CATEGORIZE THIS SORT OF A SHORT RUN OPERATIONAL INTERACTION BETWEEN COMPUTE AND POWER GRID AND EXTENDING TO THE LONGER TIMESCALE.

UH, I GUESS, YOU KNOW, NOW YOU KNOW ME WEARING THE HEAD OF PJM, UH, BOARD OF MANAGER AND, AND ER A, WE ARE SORT OF THE FELLOW TWO RTO ISOS IN THE COUNTRY THAT ARE PERHAPS MOST PRESSED FOR, FOR THIS DATA CENTER, INTERCONNECTION PROBLEMS. UH, YOU KNOW, ERCOT IS, IS SEEING, IT'S JUST INCREDIBLE AMOUNT OF, UH, INTERCONNECTION REQUESTS, UH, NUMBERS SO HIGH THAT I GUESS NONE OF US WOULD TRULY BELIEVE IN.

BUT THE QUESTION IS REALLY HOW MUCH OF UNCERTAINTY, UH, ARE WE GONNA TAKE A HAIRCUT ON, RIGHT? AND THAT, THAT, THAT IN ITSELF IS A, IS IS A BIG CHALLENGE.

IN THE MEANTIME, IN ORDER TO MEET THE DEMAND OF, OF SUCH INCREASING AMOUNT OF COMPUTE AND OTHER SOURCES OF, UH, ELECTRICITY, UH, WE, WE OUGHT TO BUILD A LOT MORE GENERATOR AS WELL.

SO BOTH ON THE DEMAND SIDE AND ON THE GENERATION SIDE, THERE'S A LONG QUEUE, UH, WAITING TO BE INTERCONNECTED.

UH, YOU KNOW, AS HARD AS WE WORK ON, YOU KNOW, REMOVING SOME OF THE BOTTLENECKS AND, AND, AND ADDING UP, YOU KNOW, 760 KV BACKBONE AND SO ON, THERE'S STILL A BIG BOTTLENECK OUT THERE ON, ON HAVING THE GRID, UH, SORT OF READY FOR THIS, UH, PARADIGM SHIFT OF, OF THE, OF THE GROWTH ON DEMAND, WHICH ALSO EVENTUALLY WILL HAVE TO BE SATISFIED BY THE GROWTH OF SUPPLY.

UM, SO THAT IS ANOTHER AREA WE SEE MANY OPPORTUNITIES AND ALSO MANY CHALLENGES THAT HAS TO BRING CLOSER OF AI AND, AND POWER.

UH, THIS PROBABLY GOES WITHOUT SAYING THAT THIS IS, ARE EXACTLY THE KINDS OF NEWS ARTICLE HEADLINES YOU SEE IN THESE STATES.

UM, YOU KNOW, YOU GO AROUND THE COUNTRY, PJM, CALIFORNIA, ER, MISO, ALL OVER THE PLACES.

UH, PEOPLE ARE GRAPPLING WITH INTEGRATING HYPERSCALERS AT SPEED AT SCALE, WHICH IS CHALLENGING.

WHAT WE ARE HOPING TO, UH, CONVEY TO YOU IN THIS PRESENTATION IS REALLY LOOKING AT THE OTHER DIRECTION, WHICH IS, HOW DO WE HARNESS THE GREAT PROGRESS IN THE AI ITSELF AND PERHAPS ADDRESS SOME OF THE BOTTLENECK ISSUES THAT WE ARE ALREADY ARTICULATING RIGHT NOW.

SO LET ME JUST FRAME THE CONVERSATION A BIT ON WHAT IS GOING ON IN THE AI WORLD.

YOU MUST HAVE HEARD ABOUT DIFFERENT NOTIONS ABOUT AI, MACHINE LEARNING, DEEP LEARNING, AND GENERAL AI.

UH, THESE TERMS ARE SOMEWHAT, YOU KNOW, FUZZY, BUT BY AND LARGE, IT, IT SPEAKS ABOUT THIS CAPABILITY, THIS INCREASING CAPABILITY THAT COMPUTERS ARE ABLE TO HARNESS BASED UPON A HUMONGOUS AMOUNT OF TRAINING DATA, AND JUST TREMENDOUS PROGRESS ON THE COMPUTE POWER, COUPLED WITH THE PROGRESS ON THE ALGORITHMS OF THE SO-CALLED TRANSFORMER ARCHITECTURE.

NOW, FOR US, THIS TRANSFORMER DOES NOT MEAN THE ELECTRICAL TRANSFORMER, UH, BUT MEANS THE ARCHITECTURE OF A PARTICULAR KIND OF NEUR NETS, UH, THAT IS VERY GOOD AT HANDLING, UH, SORT OF A SEQUENCE OF TEXT AND CODES AND SO ON.

AND THIS, AGAIN, IS NOT SOMETHING JUST HAPPENED IN THE PAST COUPLE OF YEARS.

THIS WHOLE PROGRESS OF AI HAS BEEN STARTING SOME 70 YEARS AGO, UH, AT A, A VERY, UH, UH, YOU KNOW, A, A HUMBLE BEGINNING SOMEWHERE ON NEW ENGLAND IN THE SUMMER WHEN A GROUP OF, UH, UH, YOU KNOW, TOP MINDED COMPUTER SCIENTISTS GOT TOGETHER AND, AND COINED UP THIS TERM ARTIFICIAL INTELLIGENCE.

AND, AND THE PROGRESS HAS, HAS BEEN JUST A TREMENDOUS, THERE WAS UP AND DOWNS, BUT RIGHT NOW, I THINK WE ARE SEEING ONE OF THE GREATEST, UH, SORT OF UP WAVES ON, ON, ON THE UPTAKE ON, ON AI.

ONE OF THE UNIQUE THING ABOUT THIS PARTICULAR ROUNDS OF AI IS THIS GENERATIVE

[00:15:01]

BY, BY THE WAY, THOSE OF YOU WHO USE GPT, DO YOU WHAT, DO YOU KNOW WHAT G STANDS FOR IN THE GPTG STANDS FOR GENERATIVE.

SO WHAT DOES THAT MEAN? IT IT MEANS THAT WHATEVER ANSWER YOU GET OUT OF GPT, IT IS A PROBABILISTIC COMBINATION OF PAST ANSWERS IN SOME, UH, SARCASTIC SENSE.

IN OTHER WORDS, WHATEVER YOU GET OUT OF GPT, TAKE IT WITH A GRAIN OF SALT AS GOOD AS THEY ARE.

TAKE IT WITH A GRAIN OF THOUGHT, BECAUSE ALWAYS BIAS, NATURE, GENERATIVE, RIGHT? UM, AND FAST FORWARD IN THE PAST YEAR OR SO, WHAT IS ABSOLUTELY STUNNING TO ALL OF US IS THIS PROGRESS ON AGENT AI.

SO WHAT IS AGENT AI? AGENT AI IS BASICALLY TAKING THIS LARGE LANGUAGE MODEL LIKE GPTS AND GEMINIS TO THE NEXT LEVEL TO ALLOW THEM TO INTERACT WITH THE REAL ENVIRONMENT OF A PARTICULAR WORK SPACE.

IT COULD BE IN THE POWER INDUSTRY, IT COULD BE IN THE FINANCE INDUSTRY, IT COULD BE IN EDUCATIONAL SYSTEMS IN MANY, MANY DIFFERENT, UH, SCENARIOS.

IT UNLEASHES THIS CAPABILITY OF LARGE LANGUAGE MODEL AND CONNECT THEM WITH THE WORKFLOW OF A PARTICULAR DOMAIN, RIGHT? THAT'S REALLY THE POWER OF AGENT AI.

NOW, JUST OVER THE COURSE OF PAST 12 TO 18 MONTHS, THERE HAS BEEN JUST ABSOLUTELY AMAZING PROGRESS ON AGENT AI OR AI AGENT, RIGHT? THERE ARE GENERAL ONES FOR CODING, FOR VIDEO EDITING, FOR VOICE CLONING, WHICH YOU ARE GONNA SEE A LITTLE BIT LATER.

UH, IT ALSO PENETRATES INTO DIFFERENT DOMAINS LIKE LAW, LIKE HEALTHCARE, LIKE EXPERIMENTS, UH, LIKE WHAT WE DO ON UNIVERSITY CAMPUS ON SCIENTIFIC EXPERIMENTS.

IT, IT, IT JUST UNLEASHES A LOT OF PRODUCTIVITY, UH, IMPROVEMENT.

SO THE VERY NEXT NATURAL QUESTION IS, WHAT ABOUT POWER INDUSTRY? WHERE DO WE STAND? SO IF YOU WERE TO KIND OF LAY OUT A GENERAL TIMELINE OF THE AI PROGRESS IN THE GENERAL DOMAIN VERSUS WHERE AI IS IN THE AREA OF POWER SECTOR, I THINK IT PROBABLY IS FAIR TO SAY THAT IN THE GENERAL DOMAIN, WE ARE SORT OF A, SOMEWHERE BETWEEN AGENTS AND INNOVATORS.

THAT'S WHERE WE ARE.

BUT IF YOU THINK ABOUT SORT OF A POWER DOMAIN, YOU KNOW, FOR GOOD REASONS, WE ARE KIND OF, UH, MAYBE A LITTLE BIT BEHIND, YOU KNOW, BECAUSE WE WANNA MAKE SURE THINGS ARE SAFE AND SECURE AND ALL THAT, WHICH IS ALL FOR GOOD CAUSES.

UH, WE ARE PROBABLY SOMEWHERE BETWEEN KIND OF A DOMAIN CHAT BOTS THAT'S SORT OF A, YOU KNOW, ERROR CARD OR A PJM IN-HOUSE BUILD LARGE LANGUAGE MODEL TO ADDRESS SOME OF THE SALIENT QUESTION ABOUT, YOU KNOW, WHAT IS, YOU KNOW, CRR AND SO ON AND SO FORTH INTO UNLEASHING OR COUPLING THAT CAPABILITY WITH HUMANS TOGETHER TO ADDRESS SOME OF THE WORKFLOW RELATED ISSUES.

SO I THINK WE ARE RIGHT NOW AT THE CUSP OF UNLEASHING THAT PRODUCTIVITY PROGRESS.

AND OUR GROUP KIND OF HAS BEEN, UH, PUSHING THIS, UH, OVER THE YEARS.

AND, AND THAT'S WHY WE SORT OF CAME, COME FULL CIRCLE INTO ADOPTING, UH, A AGENT AI, UH, FOR THE FIRST KIND OF, ITS PERHAPS FIRST OF ITS KIND IN INTO THE POWER DOMAIN.

UH, WHAT IT DOES IS THAT IT REALLY SORT OF, UH, PROVIDES SORT OF A, A HUMAN INTEGRATED PRODUCTIVITY IMPROVEMENT.

HERE.

THE IDEA IS NOT TO REPLACE ALL THE EXISTING TRUSTED TOOLS LIKE PSSE OR POWER WORLD THAT OR ANY OTHER ISO OR ANY OTHER UTILITY HAVE BEEN USING FOR MANY, MANY DECADES.

THAT'S NOT THE GOAL.

THE GOAL IS HOW DO WE EMPOWER THE ENGINEERS AND DECISION MAKERS IN THE HOUSE WITH THIS AGENT CAPABILITY SO THAT WHATEVER THEY THINK CAN BE DONE WITH A MUCH LESS LOW LEVEL HUMAN ENGINEER HOURS INVOLVED AND UNLEASHING A LOT MORE AUTOMATION IN THE PROCESS.

THAT'S THE CORE IDEA.

AND FOR THAT TO, UH, TO HAPPEN IN A DOMAIN LIKE POWER, IT SORT OF, UH, TAKES A MULTIPLE, UH, LAYERS COMING TOGETHER.

UH, THAT MULTIPLE LAYERS COMING TOGETHER IS THE THREE LAYERS.

UH, THE FIRST ONE IS THE FOUNDATION MODELS.

THESE ARE THE, UH, LARGE LANGUAGE MODEL THAT YOU ARE FAMILIAR WITH.

THAT IN ITSELF PROVIDES THIS ENABLING CAPABILITY FOR YOU TO REASON TO

[00:20:01]

ON TOP OF THAT, YOU WILL NEED A AGENT TOOLS.

THESE ARE THE INTERFACES BETWEEN THIS FOUNDATION MODEL AND SOFTWARE LIKE PSSE, LIKE PSK, ALL THE IN-HOUSE VENDOR GRADE SOFTWARE THAT ENGINEERS ARE DOING ON A DAILY BASIS.

AND THEN ON TOP OF THAT, IT IS A DEEP KNOW-HOW OF THE WORKFLOW, WHAT EXACTLY IS HAPPENING IN A TRANSMISSION DEPARTMENT IN ERCOT.

WHAT EXACTLY IS THE PLANNERS THINKING ABOUT WHEN THEY'RE THINKING OF DATA CENTER INTERCONNECTION REQUEST? THOSE WORKFLOW CAN ONLY BE DONE THROUGH A DEEP INTERCONNECTION OF HUMAN DECISION MAKING PROCESS AND TRANSLATING THAT INTO THIS AGE AGENT TWO, INTEGRATION ON THE SECOND LAYER.

AND THAT'S WHERE, UH, WE SPEND A LOT OF TIME WORKING ON.

I'M VERY PLEASED TO REPORT THAT, UH, WE HAVE BEEN ACTUALLY PILOT TESTING SOME OF THIS IN, UH, FELLOW ISO ENVIRONMENT AT ISON NEW ENGLAND, UH, WITH THEIR, UH, TRANSMISSION PLANNING PROCESSES.

AND, UH, GETTING A LITTLE BIT MORE INTO THE TECHNICAL DETAILS OF HOW IT ACTUALLY INVOLVES, IT INVOLVES, YOU KNOW, WE ARE SHOWING YOU THIS OPEN SOURCE VERSION OF IT, UH, WHICH IS A LARGE LANGUAGE MODEL ON TOP OF IT, EMPOWERED BY THE POWER GRID DATA SET THAT IS COLLECTED OVER THE YEARS ON MULTIPLE TIMESCALE ON MULTIPLE TASKS.

AND THESE DATA SETS WILL THEN BE USED TO FINE TUNE THE LARGE LANGUAGE MODEL THAT YOU WOULD'VE TYPICALLY GET FROM A, YOU KNOW, VANILLA VERSION OF THE LARGE LANGUAGE MODEL OFF THE SHELF, AND FINE TUNING IT TO WORK BEST IN THE CONTEXT OF POWER AND THE, UH, POWER GRID INDUSTRY.

UM, AND ESSENTIALLY IT BECOMES, INSTEAD OF THINKING OF SORT OF A SOFTWARE AS A TOOL, IT KIND OF ALMOST MOVING TOWARDS THINKING ABOUT SOFTWARE AS THIS TRUSTED PLATFORM THAT GIVES YOU THE OUTCOME YOU WOULD LIKE.

IN OTHER WORDS, YOU AS INDIVIDUAL DECISION MAKER OR ENGINEER, UM, CAN BYPASS A LOT OF THE MUNDANE, TIME CONSUMING CODING INTERFACING KIND OF TASKS SCENARIOS THAT YOU HAVE TO GENERATE.

AND THEN GO STRAIGHT TO THE KIND OF A KEY QUESTION THAT WE HUMAN BEINGS ARE MOST GOOD AT OR BEST AT, RIGHT? WHICH IS ASKING THE QUESTION, DEFINING WHAT KIND OF SCENARIO WE WOULD LIKE TO SEE, AND TRYING TO WORK SIDE BY SIDE WITH AGENT AI TO FIGURE OUT THE BEST CORRECTIVE ACTIONS.

AND, UH, IN THIS KIND OF A WORKFLOW, UH, CAN BE, UH, UTILIZED.

I, WE THINK IN A NUMBER OF, UH, SITUATIONS, UH, USE CASES IN ERCOT OR ISO OR, UH, SORT OF A POWER INDUSTRY ENVIRONMENT, WHAT IS ALSO VERY IMPORTANT TO HIGHLIGHT IS THAT IT IS NOT AI REPLACING HUMAN BEING.

IT IS NOT HUMAN BEING REPLACING AI.

IT IS HUMAN BEING AND AI WORK SIDE BY SIDE DOING THE BEST THAT THEY CAN TO SPEED UP THE PRODUCTIVITY TO REDUCE INTERCONNECTION QUEUE PROCESS TO, TO, TO UNLEASH, UH, THE GREAT THINGS THAT, YOU KNOW, WE ALL HAVE TO OFFER AS WE GO THROUGH THIS, UM, YOU KNOW, UNPRECEDENTED ENERGY TRANSITION.

SO WE WOULD LIKE TO PLAY THIS VIDEO AND, UH, FROM YOUR SITE, UH, YOU CAN PLAY IT AND, AND THEN, UH, YOU CAN JUST SEE A LITTLE BIT OF A GLIMPSE OF LIGHT OF WHAT THESE KINDS OF TOOLS CAN POSSIBLY ENABLE US.

GRID IMPACT EVALUATION IS CENTRAL TO LARGE LOAD INTERCONNECTION STUDIES, ESPECIALLY FOR DATA CENTERS.

THE WORKFLOW SPANS DATA COLLECTION, BASELINE ANALYSIS, AND POST INTERCONNECTION ASSESSMENTS UNDER NORMAL AND CONTINGENCY CONDITIONS TO VERIFY SYSTEM STABILITY AND COMPLIANCE.

IN THE ILLUSTRATED WORKFLOW, EACH AI AGENT TACKLES A DISTINCT STEP, SUCH AS RUNNING CONTINGENCY ANALYSIS IN POWER WORLD, USING DOMAIN SPECIFIC TOOLS TO STREAMLINE THE ENTIRE PROCESS.

ONCE THE USER INPUTS KEY INFORMATION, SUCH AS PROPOSED LOAD SIZE, LOCATION, AND SYSTEM MODELS, THE WORKFLOW CAN BE TRIGGERED.

EACH AGENT COLLABORATES WITHIN THE PROCESS PERFORMING TASKS SUCH AS SCENARIO GENERATION CONTINGENCY ANALYSIS, AND RESULTS VALIDATION.

AT THE FINAL STAGE, A SUMMARY AGENT COMPILES THE RESULTS INTO A USER PREFERRED REPORT FORMAT, INCLUDING ANY MITIGATION SUGGESTIONS.

IF THE INTERCONNECTION CAUSES RELIABILITY VIOLATIONS, THESE SUGGESTIONS MAY INCLUDE SYSTEM REINFORCEMENTS OR THE INTEGRATION OF SOLUTIONS LIKE ENERGY

[00:25:01]

OUR AGENTIC AI SOLUTION ALLOWS FOR FASTER, MORE MODULAR AND HUMAN INTERACTIVE EVALUATION WORKFLOWS, REDUCING ENGINEERING OVERHEAD, WHILE MAINTAINING RIGOROUS STUDY STANDARDS.

GREAT.

FIRST THING I WANNA SAY IS THAT, THAT VOICE IS NOT MY VOICE, , IT'S A SYNTHETIC VOICE.

IT IS SOMETHING THAT IS GENERATED BY AI BY JUST RECORDING 10 SECONDS OF MY ACTUAL VOICE, AND THEN THEY SYNTHETIC CREATED IT.

AND AS YOU CAN SEE, YOU KNOW, UH, IT CERTAINLY IS SHOWING A LOT OF, UH, GLIMPSE OF HOPE FOR THAT.

UH, AND YOU CAN SEE THAT THERE IS A OPEN, UH, SOURCE POWER AGENT ENVIRONMENT AT GITHUB, WHICH, UH, CHEN IS THE, UH, KEY, UH, ENABLER FOR THAT OPEN SOURCE ENVIRONMENT.

AND THERE HAS BEEN A NUMBER OF DIFFERENT USE CASES, UH, THAT IS POSTED OUT THERE.

AND WE ARE ALSO WORKING, UH, IN KIND OF A CLOSE DOMAINS WITH INDIVIDUAL ISOS SUCH AS ISO NEW ENGLAND TO DEVELOP MORE GUARDRAILS IN-HOUSE AGENT CAPABILITIES.

SO THERE ARE BOTH OPEN SOURCE MODELS AS WELL AS CLOSED SOURCE MODELS.

UH, LET ME, UH, CONTINUE ON TO COMMENT A BIT ABOUT WHAT ARE SOME OF THE CHALLENGES, UH, THAT WE SEE, UH, RIGHT NOW WITH THE, UH, DEVELOPMENT OF THIS AGENT AI CAPABILITIES.

I THINK, UM, WE ARE PROBABLY, YOU KNOW, ONE OF THE SORT OF, UH, FIRST GROUPS IN, IN THE WORLD TO, TO REALLY TACKLE THIS ISSUE IN THE CONTEXT OF POWER DOMAIN.

UH, WHAT WE LEARNED IS THAT, UH, YOU KNOW, YOU, YOU NEED TO HAVE A, YOU KNOW, VERY HUMAN IN THE LOOP PROCESS BECAUSE THIS AGENT BY THEMSELVES MAY TEND TO MAKE SOME MISTAKES WHEN THEY PERFORM TASKS END TO END.

SO FOR THOSE WHO WORRY THAT AI IS GOING TO EAT ALL OUR JOBS, AT LEAST FOR NOW, DON'T WORRY TOO MUCH BECAUSE, UH, EVENTUALLY IT WILL STILL REQUIRE HUMANS TO GET IN THE LOOP AND THEN PROVIDE SOME, UH, EH, ADJUSTMENT AND, AND FEEDBACK.

AND IN FACT, THAT'S ONE OF THE VERY FRONTIER RESEARCH TOPIC, IS HOW DO WE ENABLE THIS AGE AGENT AI TO DO REASONING AND PLANNING SIDE BY SIDE WITH HUMAN FEEDBACK, RIGHT? AND THEN THE OTHER ONE IS OF COURSE, A SAFETY AND, UH, AND PRIVACY.

WE, WE ARE LEARNING THIS AS WE SPEAK, UH, WITH THE ONGOING PROJECT WITH ISO NEW ENGLAND IS HOW DO WE OPERATE THIS IN A CLOUD SECURE ENTERPRISE CLOUD ENVIRONMENT, UH, THAT IS AMENABLE TO ALL THE REQUIREMENT ON CONFIDENTIALITY AND, AND SECURITY OF DATA, RIGHT? AND OF COURSE, YOU KNOW, DOWN THE LINE, ONE THING WE HAVE TO THINK ABOUT, UH, UH, MORE AND MORE SERIOUSLY IS HOW DO WE MAKE SURE THAT THIS KINDS OF AGENT AI IS DOING THE RIGHT THING, ETHICALLY SPEAKING, AND NOT GIVING US ANY SURPRISE, UH, WITH, UH, ADVERSARIAL UH, MINDSETS, RIGHT? SO WITH THAT, UH, I WOULD LIKE TO JUST CONCLUDE BY SAYING THAT, UH, IT IS PRECISELY BECAUSE OF THIS EXCITING OPPORTUNITY, BUT ALSO DAUNTING CHALLENGES THAT WE ARE NOW CREATING, UH, THE POWER AND AI INITIATIVE AT HARVARD SCHOOL OF ENGINEERING.

AND THE GOAL IS TO BRING THE BEST MINDS ON AI COMPUTER SCIENCE CLOSER WITH THE BEST MINDS IN POWER AND ENERGY, AND DO IT IN A WAY THAT IS INDUSTRY RELEVANT AND HELPFUL TO THE REAL WORLD.

UH, WE DO ALL KINDS OF ACTIVITIES OF CONVENING THE TECH INDUSTRY AND ENERGY INDUSTRY AND POLICY MAKERS TOGETHER.

WE HAD HONORED OF HAVING ERCOT PARTICIPATING IN ONE OF THE SYMPOSIUMS. WE ALSO HOPE TO DEVELOP TECHNOLOGIES THAT IS WELL INTEGRATING THESE TWO AREAS.

AS, AS MANY OF OF MY MIGHT, UH, REFLECT THERE, THERE SEEMS TO BE A BIT OF A CULTURAL GAP BETWEEN THE TECH INDUSTRY DATA CENTER IN PARTICULAR, AND EMPOWER INDUSTRY WHEN IT COMES TO QUESTIONS LIKE INTERCONNECTIONS.

UM, SO WITH THAT, I VERY MUCH APPRECIATE YOUR TIME AND LOOK FORWARD, UH, TO EMBARKING ON THIS AGENT AI JOURNEY WITH ALL OF YOU.

BUT REST ASSURED, WE

[00:30:01]

SO THANK YOU FOR YOUR SERVICE, AND THANK YOU FOR YOUR ATTENTION.

ARE THERE QUESTIONS FOR DR.

LECHE AT THIS POINT? I'LL ASK A QUESTION.

THANK YOU FOR YOUR PRESENTATION.

UH, YOU MENTIONED THE LARGE LANGUAGE MODEL RELATED TO INTERCONNECTION.

WHAT IS THE SECOND MOST LIKELY LANGUAGE MO LARGE LANGUAGE MODEL YOU'RE WORKING ON RELATED TO THE POWER INDUSTRY? THAT'S A GREAT QUESTION.

UM, WE, WE ARE WORKING ON A NUMBER OF ISSUES.

UH, IT'S LIKE ASKING, YOU KNOW, I HAVE THREE KIDS WHO IS MY FAVORITE KID, UH, .

BUT, UH, UH, WHAT I, WHAT I'M SEEING RIGHT NOW IS THAT FOR, FOR THE TASKS THAT ARE NOT YET IN THE REAL TIME ENVIRONMENT FOR THE TASK THAT REQUIRES A LOT OF A SCENARIO PLAY AND SIMULATION AND OFFLINE STUDY, THAT'S WHERE I SEE THE AGE AGENT AI COULD MAKE THE MOST IN WEIGHT TO AS OF NOW.

AND THERE ARE MANY OF THEM ON OPERATIONAL PLANNING, UH, ON, ON VEGETATION MANAGEMENT, ON FIRE RISKS, ON, UM, EH, YOU KNOW, STUDYING THE RELIABILITY ASPECTS OF THE GRID, A NUMBER OF SUCH ASPECTS.

YES, DR.

CHE, UH, TO FOLLOW ON THAT QUESTION, UH, I, I'M VERY INTERESTED IN THE OPPORTUNITY, MAYBE NOT FOR AGEN AI PER SE, BUT FOR MACHINE LEARNING AND OTHER AI TECHNOLOGIES TO MAKE THE WHOLE, YOU KNOW, RELIABLE DISPATCH PROCESS MORE EFFICIENT AND MORE EFFECTIVE AND, AND TO USE BETTER PREDICTIVE TOOLS, FRANKLY, IN, IN, IN THE WAY, THE WAY THAT OUR ENERGY MANAGEMENT SYSTEMS ACTUALLY WORK.

DO YOU, HAVE YOU DONE WORK ON THAT? IS THERE WORK GOING ON? DO YOU HAVE ANY COMMENTS ON THAT IDEA? THANK YOU.

UH, THAT'S A GREAT QUESTION.

I I MUST ALSO GIVE A BIG SHOUT OUT TO THE ER A TEAM THAT, UH, AS, AS THEY THINK THROUGH THIS GREAT INNOVATION, AND YOU PROBABLY WILL SEE THAT AS LATER PART OF THE AGENDA HERE.

UH, THE ER, A TEAM ACTUALLY LAID OUT A NUMBER OF THESE RESEARCH ROADMAP QUESTIONS, ONE OF WHICH IS EXACTLY ON THIS ISSUE OF ROBUST, SECURE, SORT OF A, UH, EDGE MANAGEMENT SYSTEM OR, YOU KNOW, OPTIMAL DISPATCH.

UH, THAT THERE, THAT'S AN AREA I, I SEE A LOT OF ROOM FOR OPPORTUNITY TO, TO, TO UTILIZE, UH, STATE OF THE ART AI TECHNIQUES, UH, SUCH AS REINFORCEMENT LEARNING, UH, WAYS TO BASICALLY, YOU KNOW, YOU CAN SORT OF, UH, IN YOUR MIND CONCEPTUALLY THINK ABOUT SORT OF OPERATING A GRID LIKE ERCO, ESPECIALLY FOR REAL TIME.

YOU'RE RUNNING MANY, MANY SCENARIOS AND YOU'RE TRYING TO DO A LOT OF OPTIMIZATION IN PARALLEL, AND THAT WHOLE PROCESS CAN BE LARGELY BENEFITED FROM, FROM THE PROGRESS IN AI AND MACHINE LEARNING, SUCH AS REINFORCEMENT LEARNING CAPABILITIES.

SO YES, THERE HAS BEEN WORK.

IN FACT, OUR GROUP HAS BEEN DOING SOME WORK ON THAT.

AND, YOU KNOW, AND I ALSO APPLAUD, YOU KNOW, ERIKA'S INNOVATION TEAM THINKING ALONG THOSE LINES AS WELL.

SO HOPEFULLY SOME, UH, SOME GOOD PARTNERSHIP COULD, COULD, COULD ACTUALLY HAPPEN ALONG THOSE LINES.

YEAH.

OKAY.

GREAT.

ANY, ANY OTHER QUESTIONS FOR DR.

LOS WOODY? UH, HELLO, DR.

LOSCHE, THIS IS, UH, WOODY RICKSON.

UM, DON'T HAVE NO SEE WOODY.

YES.

SO WHAT KIND OF PROGRESS IS BEING MADE AT INCORPORATING, UH, PROBABILISTIC ANALYSIS INTO SOME OF OUR, SOME OF OUR STUDIES? I MEAN, WE, WE HAVE ALWAYS FOCUSED ON POINT IN TIME AND THEN WE'VE STARTED MOVING TO PROBABILISTIC, BUT, UH, COULD AI BE A BRIDGE THAT WOULD HELP US DO MORE OF THAT? WOODY, I THINK YOUR QUESTION IS SPOT ON, AND THE SORT OF A, THE BIGGEST CHALLENGE FOR US IS REALLY THIS, HOW ARE WE GONNA DEAL WITH THIS UNCERTAINTY? AND, YOU KNOW, AND YOU GUYS ARE PROBABLY SEEING SIMILAR STORIES AS PJM IS SEEING, YOU KNOW, THIS, THIS LARGE LOAD INTERCONNECTION REQUEST COMING AND, YOU KNOW, 60 GIG OR 10 GIG, AND THEN YOU'RE LIKE, WHAT DO I DO WITH THIS 50 GIG UNCERTAINTY, RIGHT? AND THAT, THAT REALLY, UH, SORT OF, UH, POINTS TO THE ISSUE OF, YOU KNOW, YOU, YOU, YOU, YOU GOTTA DEVELOP A, A, A, A STRATEGY THAT IS SORT OF A CONTINGENT UPON SOME KIND OF A PROBABILISTIC VIEWPOINT.

LIKE, YOU KNOW, YOU'RE NOT GONNA JUST MAKE A VERY, VERY, UH, SIMPLE ANALOGY, UH, YOU KNOW, NOBODY CARES THE AVERAGE TEMPERATURE IN AMERICA, YOU KNOW, PEOPLE CARE ABOUT WHAT IS THE TEMPERATURE IN YOUR NEIGHBORHOOD, RIGHT? SO, UH, IT IS TAKING THOSE MORE GRANULAR INFORMATION, UH, IN A PROBABILISTIC SETTING, UH, THAT THAT WILL REALLY BE MEANINGFUL TO THEN JUST CONVEYING THE AVERAGE NUMBER

[00:35:01]

SO MAKING THAT IN WAY, ONE OF THE BARRIER IS COMPUTATION.

PEOPLE WOULD SAY, WELL, I CAN'T DO THAT BECAUSE THERE'S TOO MANY UNCERTAINTIES.

I DON'T KNOW HOW TO COMPUTE.

GUESS WHAT? THAT'S EXACTLY WHAT AI IS VERY GOOD AT.

SO LET'S HARNESS THE AI AND LET'S PUSH TOWARDS THAT DIRECTION.

IS THERE WORK GOING ON IN THAT REGARD AT THE MOMENT, OR IS THAT AN AREA OF FUTURE STUDY? THERE HAS BEEN WORK GOING ON IN THAT AREA, YOU KNOW, TRYING TO FIGURE OUT WHAT'S THE WAY TO, UH, HARNESS THE COMPUTATIONAL POWER OF AI TO BASICALLY ALLOW FOR A PROBABILISTIC SCENARIO PLANNING IN A MUCH MORE TIMELY MANNER.

THAT HAS BEEN SOLID WORK ON THOSE LINES.

AND, YOU KNOW, I THINK IT JUST TAKE A BIT OF A TRANSLATION AND, AND CULTURAL GAP AND, AND JUST, YOU KNOW, PARTNERSHIP BETWEEN ACADEMIA AND INDUSTRY AND POLICY MAKING TO, TO MAKE IT HAPPEN.

BUT I THINK, YOU KNOW, CHALLENGES ARE, ARE PROBABLY GOING TO LEAD US GET TOGETHER CLOSER TO EACH OTHER, SO.

OKAY.

ANY OTHER QUESTIONS? WITH THAT THEN, THANK YOU, DR.

LECHE.

UH, WE'RE GONNA MOVE ON, UH, BUT WE WILL COME BACK TO YOU WHEN WE COME INTO EXECUTIVE SESSION, IF THAT'S ALL RIGHT.

THANK YOU.

I'LL STAY ON THE LINE.

THANK YOU.

OKAY, THANKS VERY MUCH.

OKAY, THE NEXT

[5. Results of Annual Committee Self Evaluation Survey]

UM, AND AS YOU CAN SEE FROM THIS, CAN YOU SCROLL IT UP A LITTLE BIT SO PEOPLE CAN SEE THE RESULTS, NOT JUST THE HEADER.

UM, THIS, THIS IS A SUMMARY OF THE SURVEY RESULT, UM, MORE OR LESS SATISFACTORY.

THE RESULTS WERE COMPILED BY THIS, THE STAFF.

ARE THERE ANY COMMENTS? I THINK, I THINK IT'S SELF-EXPLANATORY.

UH, ARE THERE ANY COMMENTS PEOPLE WOULD LIKE TO MAKE ABOUT THIS? UM, ONE, ONE THAT I WILL MAKE WITH REGARD TO, UM, THE LAST SENTENCE ACTUALLY, AS WE, AS WE LOOK AT THE CHARTER OF THIS COMMITTEE, UH, JP AND I, UH, AND CHAD HAVE BEEN DISCUSSING THE FACT THAT AI NEEDS TO BE A MUCH MORE EXPLICIT PART OF THE CHARTER.

IT'S ALWAYS BEEN AN IMPLICIT PART, AND WE'VE HAD SEVERAL SPEAKERS ABOUT AI IN THE LAST COUPLE OF YEARS, BUT, BUT WE'RE GONNA MAKE IT EXPLICIT.

WE'RE GONNA ACTUALLY PUT IT INTO THE CHARTER.

AND I THINK ONE OF THE THINGS THAT WE CAN DO IS HELP GUIDE THE COMPANY IN TERMS OF PRIORITIES.

AS YOU'VE JUST HEARD, THERE ARE SUCH, THERE'S SUCH A BROAD RANGE OF AREAS WHERE AI COULD BE APPLICABLE.

I THINK WE SHOULD TRY AND PROVIDE SOME GUIDANCE TO, TO THE, THE REST OF THE BOARD AND TO THE WHOLE COMPANY REALLY ON WHERE TO INVEST IN AI AND WHERE, WHERE THE BEST RETURN IS.

SO THAT THAT MIGHT BE AN AREA WHERE WE CAN HAVE MORE SORT OF DIRECT INVOLVEMENT IN PRIORITY SETTING FOR THE COMPANY.

UH, ANY OTHER COMMENTS ON THAT TOO, JOHN? I WOULD, I WOULD JUST SAY THAT MAKES A LOT OF SENSE.

I THINK THAT'S THE BIGGEST CHALLENGE THAT BUSINESSES FACE IS WHAT TO PRIORITIZE ON AND ACTUALLY GET DONE.

YOU, YOU CAN TRY TO BOIL THE OCEAN IF YOU'RE NOT CAREFUL.

I, I WOULD ADD THAT PRIORITIZING IN A, WITH RTC PLUS B BEING THE UBER PRIORITY FOR THE LAST 12 TO 18 MONTHS HAS BEEN TOUGH BECAUSE WE HAVEN'T WANTED TO DISTRACT THE TEAM FROM ANY OF THE THINGS THAT THEY'VE BEEN DOING TO GET THAT DONE, WHICH WE'RE GONNA HEAR ABOUT IN JUST A MINUTE.

BUT, BUT GOING FORWARD, I THINK WE'RE GONNA GO BACK TO MORE OF A NORMAL MODEL WHERE THERE ARE QUITE A LONG LIST OF THINGS THAT ARE COMING FORWARD, AND I THINK WE CAN HAVE SOME INPUT INTO THAT.

UM, JULIE, I'M SORRY.

I WOULD, UH, YES, I LIKE YOUR IDEA.

I'M GONNA ENCOURAGE THAT WE FOCUS ON THE TECHNOLOGY ROADMAP FOR ALL ASPECTS OF ERCOT OPERATIONS, NOT JUST SOFTWARE.

AND, UM, I'D SAY THAT THE RESULT OF THAT WE COULD BE, SHOULD BE ABLE TO MEASURE IN EFFECTIVENESS, EFFICIENCIES, RELIABILITY.

SO THINK ABOUT HOW WE MEASURE THAT IMPACT AFTER THE INVESTMENT.

THE LATTER IS THE HARD PART, AS YOU KNOW, .

BUT YES, I, I GET IT.

JP, YOU, WE DO PLAN TO BRING OUR ROADMAP TO THE COMMITTEE IN THE NEXT YEAR.

SO BOTH IN EXECUTIVE SESSION AS WELL AS, UH, IN PUBLIC.

SO LOOKING FORWARD TO THAT DISCUSSION.

AND LIKE I SAID, WE ARE FREEING UP FROM RTC, SO IT'S A GREAT TIME TO GET FOCUSED ON THAT.

I THINK JULIE'S POINT THOUGH IS AS YOU BRING THE ROADMAP FORWARD, GIVE US SOME WAYS OF MEASURING, UH, OR WAYS THAT YOU ARE ASSESSING THE ROADMAP.

AND YEAH, AND PART OF THAT IS WHAT EXACTLY IS THE IMPROVEMENT WITH THAT? SO THAT'S ABSOLUTELY GOING TO BE PART OF THIS QUESTION.

GREAT.

THANK YOU VERY MUCH.

ANY OTHER QUESTIONS ABOUT THE SURVEY? OBVIOUSLY EVERY COMMITTEE HAS DONE THIS, AND THERE'S A COMMITTEE SURVEY THAT'LL BE BROUGHT TO THE FULL BOARD

[00:40:01]

OKAY.

WITH THAT, LET'S MOVE ON TO ITEM NUMBER SIX, UM, WHICH IS, UH, FEATURING THE ONE AND ONLY JP.

UM, AND STARTING WITH THE TOPIC JUJU, WHICH IS RTC PLUS B, OUR AGENDA SAYS PROJECTS, BUT, UH, WE CAN START WITH RTC BECAUSE WE HAVE A TEAM THAT, UH, HAS SOME MEETINGS.

SO MAYBE

[6.2. Real-Time Co-Optimization Implementation Update]

AND, UM, ALL RIGHT.

SO FIRST OF ALL, YES, WE ARE LIVE AND WE ARE STABLE, UM, IN THIS DECK TODAY, WHICH IS PRETTY MUCH OBSOLETE SINCE WE WENT LIVE.

I'LL ONLY COVER THE PROJECT, UH, FINANCIAL STATUS.

YEAH.

AND, BUT LATER TODAY MATT WILL COVER THE CUT OVER AS IT HAPPENED.

KEITH WILL TALK ABOUT THE MARKET IMPACTS AND DAN WILL TALK ABOUT, UH, WHAT HE'S SEEING FROM A RELIABILITY PERSPECTIVE.

UH, NOW IN MY CAREER, I FOUND IT VERY SATISFYING TO HEAR ABOUT THE SUCCESSES OF YOUR PROJECT WHEN YOUR CUSTOMERS ARE WILLING TO SHOUT ABOUT IT FROM THE ROOFTOPS.

AND HOPEFULLY I'M LOOKING FORWARD TO HEARING THAT.

BUT FOR OUR MEETING TODAY, LIKE I SAID, I'M JUST GOING TO FOCUS ON THIS, UH, ONE SLIDE, WHICH IS THE FINANCIALS, WHICH IS RELEVANT.

AND, UM, YOU KNOW, I NEVER REALLY WALKED THROUGH EVERY NUMBER HERE, BUT, UM, I THOUGHT, YOU KNOW, WE CAN LOOK BACK AND TALK ABOUT SOME OF THE SUCCESSES HERE.

OKAY, SO OUR ORIGINAL, UM, LEMME SEE IF I CAN GET MY PEN GOING.

OOPS.

OUR ORIGINAL 2019 BUDGET WAS THIS $50 MILLION.

AND AT THE CLOSE OF THIS PROJECT, WE ARE GOING TO END UP NEAR 33 MILLION.

AND I WOULD LIKE TO TALK ABOUT THAT A BIT.

OKAY, SO THIS IS NOT REALLY AN ESTIMATE, MISS.

AND IN FACT, UM, EARLIER IN, UM, 21, THERE'S EVEN IN FACT A DISCUSSION WHETHER WE NEED TO UP THIS NUMBER BECAUSE AS YOU KNOW, THIS IS A 2019 NUMBER AND COSTS ONLY GO UP.

SO WHAT HAS CHANGED, AND I'D LIKE TO SPEND, UH, SOME TALK THROUGH THIS NUMBERS.

SO WHAT HAS CHANGED IN 2020, WE NEGOTIATED AN UNLIMITED AND PERMANENT LICENSING FOR OUR DATABASE, WHICH WE LEVERAGED DURING COVID WHERE THERE WAS UNCERTAINTY IN THE MARKET, AND WE, WE KNEW THIS WORK WAS THERE.

SO WE USED THAT TO LEVERAGE THAT.

WE ALSO MADE SIGNIFICANT ARCHITECTURAL CHANGES FOR OUR DATA CENTER, WHERE WE FROM SPECIALIZED, AND THAT'S MORE EXPENSIVE HARDWARE AND OPERATING SYSTEM TO A CHEAPER, RELIABLE, MORE WIDELY USED SYSTEM.

AND THESE TWO ITEMS IS A REASON FOR, UM, ALMOST AN $8 MILLION SAVING THAT YOU SEE, UH, ON THE FIRST ROW, OKAY? WHERE, WHERE YOU'RE TALKING ABOUT HARDWARE AND SOFTWARE AND YOU CAN SEE A BIG CHUNK OF THEIR SAVING THERE.

UM, AND, AND THESE SAVINGS I REPEAT IS NOT JUST FOR RTC, IT IS A CHA SAVING THAT WE ARE GOING TO SEE IN EVERY PROJECT BECAUSE IT'S A FUNDAMENTAL CHANGE TO OUR SYSTEM.

UH, SO IT'S ALSO SOME TIMING.

WE WERE ABLE TO COMPLETE THESE PROJECTS, THESE CHANGES BEFORE RTC STARTED.

AND LASTLY, THIS PROJECT ALSO GETS CREDIT FOR HARDWARE THAT'S WRITTEN.

AND THAT'S, THAT'S THE DIFFERENCE BETWEEN, UM, THESE TWO NUMBERS HERE.

WHAT WE EXPECTED AT GATE TO EXECUTION AND WHAT WE EXPECT TO FINISH.

SO WHEN THEY RETURN THE HARDWARE, THEY WILL GET ABOUT A $2.6 MILLION CREDIT.

SO, SO FIRST OF ALL, I WANT TO THANK OUR PROCUREMENT TEAM, ARCHITECTURE TEAM AND INFRASTRUCTURE TEAM FOR THEIR FORESIGHT AND EXECUTION THAT'S RESULTED IN THIS SAVING.

NOW, IN 2021, WE PAUSED RTC BECAUSE OF WINTER STORM URI.

UH, 'CAUSE WE DID HAVE OTHER THINGS TO FOCUS ON, BUT WE DIDN'T FORCE PROGRESS ON THIS PROJECT.

WE HIRED AND RETAINED TOP TALENT, AND I REALLY APPRECIATE THE HR TEAM FOR THEIR HELP IN GETTING US THE RIGHT TALENT POOL AND HELPING US RETAIN THIS.

AND THIS TEAM DELIVERED THE EMS UPGRADE, LAUNCHED ECRS, AND THAT BUILT THE COHESION THAT THIS TEAM NEEDED.

AND YOU KNOW, WHEN, WHEN I,

[00:45:01]

UM, WE ALSO LEARNED FROM OUR EXPERIENCES, AND ONE OF THE THING WE LEARNED A LOT IS WITH OUR IMPLEMENTATION OF THE ADVANCED FAST FREQUENCY RESPONSE OR FFRA, WE HAD TO ROLL THAT IMPLEMENTATION BACK AT GO LIVE BECAUSE NOT ALL THE MARKET PARTICIPANTS WERE READY AND WE HAD TO RE-IMPLEMENT IT AFTER THEY WERE READY.

SO THAT WAS A BIG LESSON WE LEARNED ON COMMUNICATIONS AND TRACKING MARKET PARTICIPANT PROGRESS.

AND, AND THIS PROJECT SUCCESSFULLY IMPLEMENTED THAT, AND IT'S REALLY A MODEL FOR FUTURE PROJECTS.

SO A HUGE THANKS TO CLIENT SERVICES UNDER LEGAL AND FOR THEIR DIRECT OUTREACH AS WELL AS FOLKS FOR THE VARIOUS STAKEHOLDER MEETINGS.

NOW, ONE OF THE WAY THAT THE BOARD HAS HELD, AND THIS COMMITTEE IN PARTICULAR IS KEEPING THE SCOPE FIXED.

IT HAS HELD THAT MANY OF YOU ALL HAVE TECHNOLOGY BACKGROUND, SO YOU UNDERSTAND WHAT FROM IT PERSPECTIVE, WE MEAN BY KEEPING THE SCOPE FIXED AND WHAT THE SLIPPERY SLOPE IS THERE.

AND, UH, YOU KNOW, YOU KNOW, WE SAID THERE MAY BE GREAT IDEAS OUT THERE, BUT WE WON'T CONSIDER IT TO RTC UNTIL WE GO LIVE.

AND THERE'S ALWAYS NEXT, YOU KNOW, POST GO LIVE WORK.

WE CAN ALWAYS ACCOMMODATE THAT IDEA, BUT DON'T CHANGE SCOPE AND MAKE IT NEVER ENDING.

SO I REALLY APPRECIATE, UH, YOU GUYS SUPPORTING ME ON THAT.

UM, AND MY SINCERE THANKS TO KEEPING US, UH, WITH THIS, UH, SCOPE BLOCK, UM, AND NOW ALL CHANGES FROM NOW ON IS JUST ANOTHER NPRR.

WE DON'T HAVE TO WORRY ABOUT, YOU KNOW, SWITCHING HUNDREDS OF MARKET SYSTEMS AT THE SAME TIME.

AND NOW, SPEAKING OF SCOPE, I ALSO WANT TO THANK THE RTCB TASK FORCE, WHICH INCLUDES ERCOT AND MARKET PARTICIPANTS FOR THEIR WORK IN DELIVERING CLEAR AND ACTIONABLE REQUIREMENTS.

AND THAT CLARITY IS A FOUNDATION OF OUR SUCCESS.

FOR THIS, I WANT TO PARTICULARLY THANK, UH, AIRCO PARTICIPANTS, UH, S MUHI, KEN RAGSDALE, DALE MAGGIO, AND GIAN CHEN FOR THIS, FOR OUR MARKET VENDOR.

THANK YOU FOR DELIVERING THE MOST COMPLEX COMPONENT OF THIS SYSTEM ON TIME AND ON QUALITY, AND NOT JUST OUR VENDOR, BUT ALSO THE VENDORS FOR ALL THE MARKET PARTICIPANTS YOU DELIVER TOO.

AND YOUR WORK ENSURED THAT YOUR CUSTOMERS WERE READY, AND THAT MATTERED A LOT.

NOW, FOR THE PROJECT, OVER 300 PLUS PEOPLE CONTRIBUTED TO THIS PROJECT, AND THAT DOES NOT INCLUDE THE PEOPLE I JUST THANKED, RIGHT? LEGAL, PROCUREMENT, HR.

UH, SO IT IS TRULY A SYSTEM WIDE EFFORT AND ERCO WIDE ACHIEVEMENT THAT, UH, IS IN ADDITION TO THE HUNDREDS OF MARKET PARTICIPANTS WHO WORKED WITH US.

UM, SO THIS IS FROM MY PERSPECTIVE, WHAT PABLO TALKS ABOUT AS ERCO 4.2.

SO NOW I WOULD LIKE TO TALK ABOUT THESE PROJECTS AND LIKE TO INVITE THE LEADERS OF THESE PROJECTS TO JOIN ME ON STAGE, AS I CALL OUT THEIR NAMES.

UH, THESE ARE JUST THE LEADERS, WHAT WHOEVER CALL OUT IS THE PROJECT MANAGER, THE BUSINESS OWNER, AND THE TECHNICAL LEAD FOR EACH OF THESE PROJECTS.

SO I LIKE THAT TO ACKNOWLEDGE THE FEW PEOPLE WHO HAVE DELIVERED THIS PROJECT WITH, UH, 34% UNDER BUDGET AND, UH, SIX MONTHS AHEAD OF SCHEDULE.

UM, SO, SO I'M GOING TO CALL OUT, UH, THE GMS PROJECT.

THE BUSINESS OWNER IS MATT , WHICH YOU, YOU GUYS ARE FAMILIAR WITH.

MATT.

UM, PLEASE COME ON HERE SO THAT, UH, PEOPLE IN PUBLIC CAN SEE YOU AS WELL.

UM, UH, THE TECHNICAL, UH, LEADER, SHRUTI, HARI, HARAN, AND W PURI, AND THE PROJECT MANAGER IS LEO CASTILLO FOR, UH, THE SHRUTI, WA, AND LEO FOR COMMERCIAL LABS.

THE BUSINESS OWNER OWNER IS AUSTIN ROSEL, THE TECH LEAD IS PARIS THE WAY, UM, AND THE PROJECT MANAGER IS OSA O EQUA FOR INTEGRATION TEAM.

THAT TEAM COULDN'T BE HERE BECAUSE THEY ACTUALLY HAVE DEPLOYMENTS GOING ON, AND IT'S A REMINDER THAT, UH, THE RTC PROJECT IS NOT OVER AND REALLY RUNS THROUGH JUNE, BUT I'LL CALL OUT THEIR NAMES.

THE BUSINESS OWNER IS JOHN MESSER, THE TECHNICAL LEADERS, ANGELA MICKELSON.

AND THE PROJECT MANAGER, AGAIN, IS ALSO

[00:50:01]

RIN WAS, WAS FOR THE QUARTERBACK FOR THE CUTOVER.

AND LASTLY, AND NOT LEAST, FOR THE OVERALL PROJECT, WHICH IS THE FIRST LINE, UM, THE BUSINESS OWNER AGAIN, IS MATT, IN HIS DOUBLE ROLE, THE TECHNICAL LEAD IS SATIA, KRISHNAN, UH, OVERALL TECHNICAL LEAD, AND THE OVERALL PROJECT MANAGER IS JERRY.