A living AI roadmap for a national gas network

When the Australian Gas Infrastructure Group (AGIG) set out to move from AI ambition to AI in practice, they needed an honest read on:

• AI readiness and 
• A prioritised initiative demand pipeline they could trust to deliver value as quickly as possible. 

Over a few months, Hapsicle deployed an AI Director that, when prompted, is autonomous and manages various other agentds to complete a outcome it is tasked to complete. Hapsicle & the AI Director proposed and worked with AGIG to run two linked streams of work: 

• an AI readiness assessment and
• an AI Lab that turned the team's demand into a prioritised portfolio of 56 costed use cases.

Once the AI maturity assessment was completed and the demand pipeline agreed, the AI Director, (subsequently called "Brad") merged the streams to prioritise AGIG's ability to deliver based upon its current AI maturity. 

The outcome was reviewed and as the demand cases shifted or maturity was reviewed, the roadmap rebalanced itself and foundational investments like data quality are prioritised by the customer outcomes they unlock. A true end-to-end uplift linking customer benefits / outcomes to foundational Data & AI investments

...managed and deployed another 59 agents to: 

• Listen to and update progress based upon 26 hours of meetings
• read, analyse and incorporate 14 documents into its learning
• complete 45 in-depth interviews 
• develop 56 prioritised, use cases
• create 11 methodology frameworks 
• create 8 dimensions of AI maturity then graded the organisations maturity with that framework
• develop a prioritised demand pipeline that weighted the organisations maturity (i.e. the ability to deliver) in its assessment
• deliver thorough and detailed audit trails when asked
• learn. The proof? Brad inferred the overall strategic direction (that was then double checked by a human) from all those documents, meetings and interviews to help in assessing the demand prioritisation. 

Yes, Brad, the Autonomous AI Director, devised and wrote a strategy with no direct human prompt or uploaded document (or other input), through nothing more than listening & learning on this assignment for 12 weeks. 

AGIG wanted to move from talking about AI to doing it — without a scatter of disconnected pilots or a strategy written in a room. They needed two things at once: an honest, evidence-based read on how ready the organisation really was, and a prioritised pipeline of AI opportunities they could trust. Run the conventional way — strategy, then use cases,

then adoption — that sequence takes 12–18 months before anyone sees a benefit.

Rather than sequence the work, Hapsicle ran two parallel streams that shared a single engine.

Stream 1 — AI Adoption Readiness. A deep assessment across eight dimensions. The baseline of the strategy - the current state, was built from evidence, not assumption.

Stream 2 — AI Lab. A delivery framework that turned the demand AGIG's teams were already voicing into a prioritised portfolio of 56 real use cases with measurable outcomes. 

Hands-on learning ran through both streams: every interview built AI literacy, every workshop showed the team what “good” looked like.

Most organisations run strategy and delivery as separate tracks that meet at quarterly  checkpoints. Hapsicle wired the two streams together that remained constantly intertwined. 

Every use case was weighted not just by business value, but by AGIG's actual ability to deliver it given its data and organisational maturity. A high-value idea that depended on foundations AGIG didn't yet have was sequenced, not discarded.

When AGIG's own data team supplied a detailed data-quality assessment, the weighting updated in minutes and the portfolio reshaped itself — no rework, no three-month consulting cycle. 

 This program solved a problem that regularly stalls AI investments: the Data uplift gap between the technology teams who need to invest in Data Management foundations and the business teams who want customer outcomes. Because maturity and pipeline were linked, foundational investments — Data Governance, integration, quality — could be prioritised by the outcomes they unlock. Data Governance wasn't “a cost now for a vague payoff later”; it was the thing that unlocked three high priority use cases that could achieve value and customer improvement now as quick wins. 

The engine behind the readiness assessment was Brad — an autonomous, conversational AI interviewer Hapsicle configured for the engagement. Brad conducted 43 interviews across the Customer & Community division and adjacent functions — IT, Legal, Finance, Governance, Data and Strategy — each 30–60 minutes, covering all eight dimensions, each following its own path. Brad scheduled himself, ran the conversations in real time, saved progress, and produced structured, searchable transcripts ready for analysis. The methodology, scoring and strategic framing were all Hapsicle executive input which came from built on 30+ years experience in COO / CDO roles in regulated industries.

Hapsicle is a human lead, AI powered consultancy that has been built from scratch with autonomous AI at its core. Today, Hapsicle has ~50 autonomous AI Directors (that manage hundreds of subordinate agents) that take their input (and answer to the governance!) of seasoned industry Hapsicle humans who instruct the Directors.  To understand the balance of work, for example, the humans:

• devised the two-stream approach, 
• set the structure for 11 foundational frameworks (e.g. the 8 AI readiness assessment criteria was one framework) that drove how the program would be executed
• rigorously reviewed, shaped and challenged Brad's outputs, and
• generally treaded Brad (and other AI they used) as a midpoint, (i.e. to be checked by a human) rather than an endpoint, (i.e. giving final Brad results directly to AGIG for decisioning)

• Conversation, not a form — 15–40 follow-up questions per person, each interview its own path.
• Adaptive probing — when something surprising surfaced, Brad followed it, sometimes six layers deep.
• Real-time reframing — when a question didn't fit a role, he pivoted on the spot instead of collecting “N/A”.
• Read the room — adjusted to five-minute executives and hour-long enthusiasts alike: 90% completion, versus the 30–50% typical of internal surveys.
• Tone and sentiment — frustration, enthusiasm, resignation all registered, and shaped the scoring and approach to questioning.
• Never lost a thread — progress saved automatically; senior stakeholders finished across multiple sessions.
• Spoke AGIG's language — aligned to AGIG's Human Synergistics culture framework, so findings landed as growth opportunities, not blame.

In a few months, AGIG had the start of a strategy grounded in real evidence, a prioritised pipeline of 56 costed use cases, (including quick wins), and a workforce that had genuinely worked alongside AI and been shaped by it. Most importantly, AGIG had:

• a living roadmap — one that keeps sequencing investment in the right order as the evidence changes, with the confidence to defend every call.
• "Relief" that that were finally doing something productive with AI
• "Confidence" to do more and go faster