Process Discovery · Co-Design Alignment

From working product
to first customers.

Five decisions turn the product we built into its first real customers — one box per block, filled in this room.

90 minutes · five decisions · → we leave aligned.

Process Discovery Workshop · 2026-07 | Guided by Aqib · Right-customers block by the ANZ team

The contract — five boxes, five decisions.

1North star→ one-liner agreed — the sentence we all say to a customer✓

2Reality→ MVP promise — we deliver X, explicitly not Y✓

3Co-design→ scope + goals + what we need from the client✓

4Right customers→ ICP rules + named roster · ANZ presents✓

5Operating plan→ structure ratified + named owners✓

We're not reviewing slides. We're filling 5 boxes.

Disagreement is the point — say it in the room, we capture it on the slide, and every box ends ✓ decided or ⏸︎ parked with an owner.

1

Block 1

North star

decision to reach ▸ the one-liner we all say (box 1)

1north star

2reality

3co-design

4customers

5operating

?

sales“documentation”

product“an ontology”

ANZ“a cockpit”

eng“interviews”

exec“a runbook maker”

Ask five of us what this product is — you get five answers. That's five different pitches to customers.

▸◂one sentence we can all say (box 1)

Vision

“We come from the automation, not the process.”

① MAP

All processes in the company

why: do they exist & which are worth solving?

how: interviews → the digital twin → opportunity cockpit

② UNDERSTAND

ONE process, all flows

why: set up the agent flows

how: campaign + interviews → runbook → agent on Beam

③ OPTIMIZE

Real work data for accuracy

why: raise agent accuracy

how: task mining → ground truth → agent learning

find the work ──▶ learn the work ──▶ perfect the work

Three jobs — find, learn, perfect. Lead with the intelligence, never with the agents.

“Process discovery creates a digital twin of your company — and makes it easy to turn its hidden potential into real, running agents.”

twin = MAP ▤ potential = ranked ▶ running = ship

alternates · tap to swap

discuss ▸ does it hold all 3 pillars? · can a customer repeat it back? · does “easy” ring true?

transition ▸ “That's the direction. Honestly — where are we on the road?”

2

Block 2

Reality

decision to reach ▸ MVP sufficient? — deliver X / not Y (box 2)

1north star

2reality

3co-design

4customers

5operating

Half of us have seen…

● REC live interview — “walk me through it…”

…the other half have seen

◉ demo cockpit — constellation + ranked bubbles

→ We're about to promise different things to customers.

decision to reach ▸ is the MVP scope sufficient? — deliver X / not Y (box 2)

What we have today — 4 building blocks

36 issues live

① Campaign creation

create · invite · context upload

● LIVE

▶

② Voice interviews

Discovery interview insights in Beam Discovery

link-based · end-to-end · knows who it talks to

● LIVE

▶

③ Task mining

share screen, do the work → traces

◐ TUNING

▶

④ Outputs

SOP · flowchart · process report · download

● LIVE

○ also there, on demo data ▸ontology graph · opportunity matrix Ontology constellation

● foundation ▸discovery.beam.ai · SOC 2 · auth · auto-deploy · design system · workspaces v1 · copilot v0

Nothing on this slide is a promise — everything is clickable now.

What's being added

This week

Campaign setup rebuildgoal · people · context

Workspaces + MCPisolated — every query needs a workspace ID

Next week

Runbook showcase

Internal test~1 wk, dogfood our own HR processes

the fold

Beyond · not promised

live ROI cockpit

Now / Next / Later

chat-driven setup

knowledge drop-in

The question ▸ is this scope sufficient to do process discovery with a real customer — not “is it finished”.

deep understanding of how their existing processes actually work

a runbook → lets them create the agent flow

a process report per campaign — process well described

MINIMUM BAR ▸ enough to run discovery with a real customer

below the bar · not promised ▸ ✗ live ontology ✗ live cockpit ✗ day-1 polish ✗ custom features

discuss ▸ what's missing that would block a partner? · would we run this on OUR process this week? · what will a partner complain about first?
decide ▸ scope sufficient: YES / YES-IF

transition ▸ “The gap closes fastest with real customers in the loop. So — what IS co-design?”

3

Block 3

Co-design

decision to reach ▸ scope + goals + needs (box 3)

1north star

2reality

3co-design

4customers

5operating

DESIGN PARTNERSHIP

“co-development”

“a product trial”

“free consulting”

“custom build”

Customers hear whichever version their rep believes.

decision to reach ▸ one scope + goals we all sell (box 3)

Co-design IS / IS NOT

✓ It is

✓Early access to the whole product — use it, embed it
✓Learning agentic process discovery as a new way of working
✓Their feedback shaping the roadmap every two weeks

✗ It is not

✗Co-development — no custom builds, no client dev/product teams in our codebase
✗On-premise deployment
✗A consulting audit
✗A finished-product trial those wait for group 2

“We give them the product early. We don't build them a product.”

The loop — per partner

Weeks 1–3 · learn the work

Weeks 4–6 · make it actionable

AI voice interview10–15 min each

debrief w/ intervieweeright after each interview

× ~10 · 10–15′
all in 3 wks

(Aqib + Lucas run debriefs)

PROCESS REPORT

process owner reviewson the outputs

iterate outputsuntil actionable

weekly

exit ▸ ✓ ACTIONABLE

value bar ▸ every interview → a confirmed process draft.

IN SCOPE

early product access · ~10 discovery interviews · debriefs · process report · roadmap input

✗ custom development ✗ integrations ✗ SLAs ✗ on-premise

goals ▸ 1 interview quality 2 find where outputs break 3 logos + customers who want to buy

We need — per partner

◉ 1 exec championkickoff + W6 readout

◉ 1 process ownerkickoff · ~W3 mid-mark · weekly W4–6 on outputs

◉ 2–3 doers~10 intvws × 10–15′ + debrief right after ≈ 3 HOURS total

discuss ▸ in/out of scope we disagree on? · is a 3-hour work-level ask realistic here?

transition ▸ “This only teaches us if the customer is the right kind.”

4

Block 4

Right customers

decision to reach ▸ ICP rules + named roster (box 4)

1north star

2reality

3co-design

4customers

5operating

4

Block 4 · ▸ problem · ANZ presents

Right customers

1north star

2reality

3co-design

4customers

5operating

The wrong first partner poisons the loop.

decision to reach ▸ ICP rules + named roster (box 4) · note ▸ co-design FILTER, not the full sales ICP

The exchange

They get

Cutting-edge process discovery, hands-on
ONE process deeply understood — flows, edge cases, handoffs
Team trained on the product + the new way of working
A free introduction — test before committing

We get

Logos we can sell in sales calls
Use-case feedback → a more user-friendly product
Potential customers who like it and want to use it

boundary ▸ we do NOT build custom features

FIT ▸ ✓ knows which process · ✓ building agents · ✓ SaaS-ok · ✓ expectation-tolerant · ✓ starts mid-Aug

↪ full ICP (mid-market $250M–$2B + BPO) = post-MVP

① ICP CRITERIA as filtered ▸ edits

② EXCEPTION RULE more money, misses fit — bend? money ≠ right-kind feedback decision

③ ROSTERstar the first pair · cap ▸ see block 5

AmeripriseUS · F230 · task mining

ComdirectDACH · bank · POC

LMGANZ · front-runner

WisrANZ

transition ▸ “Names on the board. Now the machine.”

5

Block 5

Operating plan

decision to reach ▸ structure ratified + named owners (box 5)

1north star

2reality

3co-design

4customers

5operating

5

Block 5 · ▸ problem

Operating plan

1north star

2reality

3co-design

4customers

5operating

~60 interviews + a debrief after each. All at once ▸ same feedback ×6 · two people drown.

decision to reach ▸ structure ratified + named owners (box 5)

The cohort machine

max 6 partners · onboarded in pairs · every 2 weeks

why pairs, why staggered ▸ take feedback from two, implement, next pair starts on a BETTER product. All-at-once = the same feedback six times.

The 6-week engagement — per partner

retires the old 8-week design-partner collateral — confirm w/ ANZ.

6 wks · max 6 partners · pairs every 2 wks · fixed milestones, flexible interiors

Ratify

Product engineeringAAqib

Solution eng + agent promptLLucas

DebriefsAAqibLLucas

Exec & process-owner mtgsGGTM attends

Client side◉1 champion◉1 owner◉2–3 doers≈ 3h total

discuss ▸ Lucas's capacity? · cohort-1 start Aug 15 confirm

transition ▸ “Five boxes. Let's read them back.”

Read-back — the contract, filled.

1North starpending

2Realitypending

3Co-designpending

4Rosterpending

5Operatingpending

⏸︎ Parked item owner date

From working productto first customers.

The contract — five boxes, five decisions.

North star

“We come from the automation, not the process.”

Reality

Half of us have seen…

…the other half have seen

What we have today — 4 building blocks

What's being added

Co-design

Co-design IS / IS NOT

✓ It is

✗ It is not

The loop — per partner

We need — per partner

Right customers

Right customers

The exchange

They get

We get

Operating plan

Operating plan

The cohort machine

The 6-week engagement — per partner

Read-back — the contract, filled.

From working product
to first customers.