This Part­ner Insight was authored by Phil Schwarz, Senior Indus­try Strate­gist for Ener­gy and Resources at Open­Text.

Util­i­ties are nav­i­gat­ing one of the most sig­nif­i­cant peri­ods of trans­for­ma­tion in their his­to­ry. Grid mod­ern­iza­tion, renew­able inte­gra­tion, elec­tri­fi­ca­tion, work­force turnover, and increas­ing ener­gy demand are cre­at­ing new oper­a­tional and finan­cial pres­sures. At the same time, orga­ni­za­tions are explor­ing how AI assis­tants, copi­lots, and autonomous agents can help improve pro­duc­tiv­i­ty, accel­er­ate deci­sion-mak­ing, and stream­line busi­ness processes.

As SAP cus­tomers look to oper­a­tional­ize AI, one real­i­ty is becom­ing clear: suc­cess depends less on the AI mod­el itself than on the infor­ma­tion foun­da­tion beneath it. Every AI agent ulti­mate­ly needs trust­ed infor­ma­tion to make deci­sions, gov­erned author­i­ty to act, and secure mech­a­nisms to exchange infor­ma­tion across orga­ni­za­tion­al boundaries.

This is where Open­Text Solu­tions for SAP play a crit­i­cal role. Open­Text helps orga­ni­za­tions con­nect con­tent, process­es, records, and engi­neer­ing infor­ma­tion direct­ly to SAP busi­ness trans­ac­tions, cre­at­ing a gov­erned infor­ma­tion foun­da­tion that enables AI to scale safe­ly and effec­tive­ly across the enterprise.

As util­i­ties move from AI exper­i­men­ta­tion to pro­duc­tion deploy­ment, three capa­bil­i­ties define what every AI agent requires to oper­ate with­in a reg­u­lat­ed, mis­sion-crit­i­cal envi­ron­ment: Decide, Act, and Transact.

Decide: Read Trust­ed Context

Before an AI agent can make a rec­om­men­da­tion, it must first under­stand the sit­u­a­tion it is eval­u­at­ing. That requires access to trust­ed context.

In util­i­ty oper­a­tions, deci­sions depend on engi­neer­ing draw­ings, main­te­nance pro­ce­dures, inspec­tion reports, work his­to­ries, asset doc­u­men­ta­tion, com­pli­ance records, oper­a­tional reports, cus­tomer bills, and count­less oth­er sources. Yet these records often reside across mul­ti­ple sys­tems and repos­i­to­ries, mak­ing it dif­fi­cult to ensure employ­ees — and AI agents — are work­ing from com­plete and cur­rent information.

The chal­lenge grows more acute as orga­ni­za­tions deploy AI. Agents can only make deci­sions based on the infor­ma­tion avail­able to them. If infor­ma­tion is incom­plete, out­dat­ed, or dis­con­nect­ed from busi­ness process­es, the result can be inac­cu­rate rec­om­men­da­tions deliv­ered with unwar­rant­ed confidence.

The answer is gov­erned content.

The agent reads gov­erned asset and oper­a­tional records that are cur­rent, com­plete, and prop­er­ly orga­nized accord­ing to busi­ness, reg­u­la­to­ry, and oper­a­tional requirements.

This is where Open­Text Con­tent Man­age­ment becomes foun­da­tion­al. By con­nect­ing enter­prise con­tent direct­ly to SAP busi­ness process­es, util­i­ties cre­ate a trust­ed infor­ma­tion envi­ron­ment where doc­u­ments, records, engi­neer­ing infor­ma­tion, and oper­a­tional con­tent stay gov­erned through­out their life cycle.

When AI agents access con­tent through a gov­erned infor­ma­tion lay­er, they work from approved ver­sions, val­i­dat­ed records, and con­trolled sources. Main­te­nance teams gain access to cur­rent pro­ce­dures. Engi­neers work from approved doc­u­men­ta­tion. Field ser­vice per­son­nel can quick­ly locate the infor­ma­tion required to sup­port crit­i­cal decisions.

In an AI-enabled util­i­ty, trust­ed infor­ma­tion is not option­al — it is the foun­da­tion for every sub­se­quent decision.

Act: Ensure AI Oper­ates With­in Autho­rized Boundaries

Once an agent can access trust­ed infor­ma­tion, the next ques­tion is whether it should act on that information.

For util­i­ties, gov­er­nance is the dif­fer­ence between an AI pilot and a pro­duc­tion-ready AI capa­bil­i­ty. Every orga­ni­za­tion must answer a sim­ple but crit­i­cal ques­tion: What is the agent allowed to do?

The prin­ci­ple is straight­for­ward: The agent does only what you grant­ed — and every action is logged.

Util­i­ty busi­ness process­es require approval con­trols, seg­re­ga­tion of duties, auditabil­i­ty, com­pli­ance over­sight, and doc­u­ment­ed account­abil­i­ty. AI can­not oper­ate out­side these require­ments. Instead, it must func­tion with­in the same gov­er­nance frame­work that guides human workers.

A strong exam­ple is Open­Text Ven­dor Invoice Man­age­ment for SAP Solu­tions. Invoice pro­cess­ing is a process where orga­ni­za­tions want greater speed and effi­cien­cy with­out sac­ri­fic­ing con­trol. Intel­li­gent automa­tion can clas­si­fy invoic­es, extract infor­ma­tion, val­i­date data, route excep­tions, and sup­port work­flow exe­cu­tion. How­ev­er, the sys­tem remains gov­erned by estab­lished approval process­es, busi­ness rules, autho­riza­tion con­trols, and audit requirements.

Every action remains vis­i­ble. Every excep­tion remains trace­able. Every approval remains documented.

Trans­act: Give AI a Trans­ac­tion Passport

Many of the high­est-val­ue util­i­ty process­es extend beyond com­pa­ny bound­aries. Cap­i­tal projects involve engi­neer­ing firms and con­trac­tors, pro­cure­ment involves sup­pli­ers and man­u­fac­tur­ers, and com­mer­cial oper­a­tions involve cus­tomers and exter­nal partners.

As AI evolves, agents will increas­ing­ly par­tic­i­pate in these exchanges.

To do that, they need what Open­Text describes as a trans­ac­tion pass­port — the abil­i­ty to exchange infor­ma­tion across orga­ni­za­tion­al bound­aries through iden­ti­ty, trust, and ver­i­fi­able evidence.

A trans­ac­tion pass­port answers sev­er­al crit­i­cal questions:

• Who is par­tic­i­pat­ing in the transaction?
• What infor­ma­tion is being exchanged?
• Can the infor­ma­tion be trusted?
• Is there a ver­i­fi­able record of what occurred?
• Can the orga­ni­za­tion prove com­pli­ance lat­er if challenged?

For util­i­ties, one of the clear­est exam­ples is engi­neer­ing col­lab­o­ra­tion dur­ing cap­i­tal projects.

Engi­neer­ing draw­ings, spec­i­fi­ca­tions, design doc­u­ments, red­lines, and tech­ni­cal pack­ages rou­tine­ly move between util­i­ties, EPC firms, con­trac­tors, and sup­pli­ers. With­out gov­er­nance, these exchanges can cre­ate delays, ver­sion-con­trol prob­lems, rework, and com­pli­ance risks.

Solu­tions such as Open­Text Con­tent Man­age­ment for Engi­neer­ing help address this chal­lenge through capa­bil­i­ties includ­ing trans­mit­tal and sub­mit­tal man­age­ment. These process­es gov­ern the exchange of project doc­u­men­ta­tion while main­tain­ing trace­abil­i­ty, ver­sion con­trol, review work­flows, approvals, and audit trails.

When an engi­neer­ing doc­u­ment is shared, stake­hold­ers can ver­i­fy which ver­sion was sent, who reviewed it, what changes were request­ed, when approvals occurred, and what doc­u­men­ta­tion sup­ports the final deci­sion. The entire exchange becomes gov­erned and defensible.

This same prin­ci­ple applies to the future of agen­tic AI. Before an AI agent can safe­ly engage with sup­pli­ers, con­trac­tors, part­ners, or com­mer­cial cus­tomers, it must have trust­ed iden­ti­ty, gov­erned infor­ma­tion, and ver­i­fi­able evi­dence. In oth­er words, it needs a trans­ac­tion passport.

Build­ing a Gov­erned Foun­da­tion for AI-Ready Utilities

Every effec­tive AI agent depends on three capa­bil­i­ties. It must read trust­ed con­text to decide, oper­ate with­in autho­rized bound­aries to act, and hold a trans­ac­tion pass­port to trans­act across orga­ni­za­tion­al bound­aries through iden­ti­ty, trust, and ver­i­fi­able evidence.

Through its port­fo­lio of SAP-inte­grat­ed infor­ma­tion man­age­ment solu­tions and its long-stand­ing SAP part­ner­ship, Open­Text helps util­i­ties build the gov­erned foun­da­tion that con­nects con­tent, records, engi­neer­ing infor­ma­tion, work­flows, and busi­ness process­es direct­ly to SAP environments.

Ready to learn how Open­Text can help your util­i­ty estab­lish an AI-ready infor­ma­tion foun­da­tion for SAP? Speak to an expert.

Phil Schwarz is Senior Indus­try Strate­gist for Ener­gy and Resources at Open­Text.