B2B SaaS and workflow software

The happy path is the easy part.

B2B software becomes difficult where work changes hands: one record, several roles, conditional authority, exceptions that matter and a history that must still be trusted afterwards.

When workflow software starts to fight the work

Tcules designs and builds around the operating reality of B2B software. This is especially relevant after product-market fit, when one successful workflow has become many variants across teams, plans, markets and integrations.

The same record means different things to different roles

Sales sees an opportunity, operations sees a commitment and support sees an exception. The shared history cannot reliably connect them.

Status names must connect to what can happen next

A label looks clear until two roles infer different ownership and permitted actions from it.

Permissions have grown as one-off exceptions

The access matrix says one thing; escalations, delegation and regional rules say another.

People export data to make the real decision

The product stores the record but does not provide the comparison, context or action the work requires.

Automation returns exceptions to chat and spreadsheets

The normal route is fast; the product abandons the work as soon as the rule does not fit.

AI has been added without an authority model

A suggestion, draft and action appear in the same visual language even though they carry different consequences.

Model the workflow before redesigning it

  1. 01

    What persists?

    Requests, accounts, assets, campaigns, incidents, projects or meetings remain while their presentation changes.

  2. 02

    Who may see and change it?

    Roles carry responsibility, authority and hand-off as well as access.

  3. 03

    Which rules move it?

    Eligibility, assignment, validation, approval and automation govern state changes.

  4. 04

    Which exceptions matter?

    The product becomes dependable when it explains and recovers work outside the normal route.

  5. 05

    Which decision must the product support?

    Dashboards, alerts and AI recommendations matter only when a person can act with sufficient basis.

The workflow design and information architecture definitions clarify the difference between screen flow and the objects, rules and states underneath it.

Test the surface hypothesis against the product structure

What shallow redesigns miss
Visible symptomInitial interface hypothesisProduct question Tcules asks

Too many form fields

Test whether fields can be removed or combined

Which later decisions depend on each value, and who can safely supply it?

Complex approval flow

Test whether steps can be collapsed

Which controls are essential, where does authority change and how is disagreement recorded?

Inconsistent status labels

Test whether copy can be standardised

Do the underlying states and permitted transitions actually mean the same thing?

Dashboard overload

Test whether cards can be reduced

Which role is deciding what, from which source, with what uncertainty and next action?

AI assistant

Test whether a conversational entry point helps

What becomes durable state, what may the model change and how does a person recover?

Relevant work

Priority inside scheduling

Doodle

The public case connects meeting type, organisational priority, tentative availability and bookability confidence. It documents priority and scheduling-object modelling without claiming to expose the complete bookability model.

Read the Doodle case

A rule before creation

Dealpath

The public case narrative records a search-before-create rule and distinguishes the original listing context from the attach-versus-create branch.

Read the Dealpath case

Review authority around AI

BuildTwin

The published review surface shows AI-assisted and manual checks, per-check status and issue raising around a drawing.

Read the BuildTwin case

A transaction becomes operational work

Gold Image Printing

The product connected customer ordering with artwork, production and fulfilment work that could not be redesigned as isolated screens.

Read the Gold Image Printing case

How AI changes the workflow

AI can retrieve material, summarise state, recommend a next action or execute a bounded step. Those are different authority levels.

A recommendation needs basis and alternatives; an action needs permission, preview, durable state and recovery. Probabilistic work should not look like a deterministic status merely because both fit inside the same table.

Implementation still matters

  • Identity and permissions across services
  • Event history and idempotent changes
  • Latency, partial success and interrupted work
  • Integrations that disagree about current state
  • Accessible review and correction paths

When this expertise is useful

This work fits an active B2B product initiative with meaningful workflow stakes and a team willing to examine the records, rules and authority beneath the interface.

It can begin with Product Design, a UX Product Audit, Modernisation or a defined software slice.

Where Tcules is the wrong choice

  • Rate-led production outsourcing
  • A request to decorate screens while workflow behaviour is out of scope
  • No access to people or materials that explain the operating work
  • A requirement for Tcules to supply unverified domain or regulatory policy

Choosing a starting point

  1. 01

    Product Design, audit, modernisation or a software slice?

    • Choose Product Design when the workflow model or behaviour needs definition.
    • Choose an audit when the cause is disputed.
    • Choose Modernisation when learned work and transition matter.
    • Choose software delivery when the product choice is supported and Tcules must own the agreed implementation.
  2. 02

    What will the team receive?

    The output follows the starting route: a product model and interaction specification, bounded findings and recommendations, a modernisation sequence, or working software with agreed release evidence.

  3. 03

    Who and what need to be available?

    Bring representative roles, records, exceptions and product access, plus people who understand domain policy, customer-facing work, product choices and implementation constraints.

Tell us where the workflow breaks down

Share one workflow that keeps moving into chat or spreadsheets and the roles that see it differently.