How Connect writes faster, stronger proposals with an A.I. sales assistant

Project Overview

Who is Connect?

Connect is a communications agency specialized in environmental and infrastructure communication — think communication around construction projects, participation processes and stakeholder management. The agency has around sixty employees and operates from Belgium and the Netherlands, with teams for advice, creative work, operations and HR. Many assignments come in through public tenders, which means writing a large number of proposals and plans of approach every year, each based on an extensive brief.

Connect already used A.I. in daily work, but wanted to go further than loose tools. The question was: where's the real win for the whole organization?

The challenge

"How do we use A.I. in a way that saves time, raises the quality of our work, and gets buy-in from the whole team?"

Connect didn't want surface-level A.I. applications — it wanted a well-founded choice. First find out where the value is, then build.

Project Delivery

The approach

The project ran in three steps. First we mapped every A.I. opportunity, then we worked out one opportunity into a business case, and only then did we build.

1 · A.I. Opportunity Mapping

We started with an A.I. Opportunity Mapping. In a half-day workshop with a team from strategy, management, communications, creative, HR and admin, we mapped every A.I. opportunity, department by department.

We first gathered pain points and ideas without judging them. Then we scored each idea on value and complexity, which produced a clear picture: quick wins that can start right away, strategic initiatives worth building, and a backlog for later.

That exercise surfaced four big opportunities, with knowledge sharing as the common thread. Employees lose time hunting down past proposals, plans and cases. The estimated win from a good knowledge assistant alone was about one hour per person per week.

The benefit: management ended up with a shared roadmap instead of loose ideas.

2 · Business Value workshop

Next we zoomed in on one opportunity with a Business Value workshop. During that session we realized together that a general knowledge database was too broad to deliver real value. We shifted the focus to where the win is biggest: writing proposals.

In the workshop we mapped the current proposal process, sketched out the desired process, identified the users and their requirements, and prioritized those using the MoSCoW method. We also set measurable goals: less time per proposal, more proposals per year, more reuse of strong material, and a higher score on the agency's own quality criteria.

The benefit: a concrete scope and a business case, with an MVP small enough to deliver fast.

3 · Sales Buddy in production

Next we built Sales Buddy, Connect's sales assistant. We worked in three short sprints, with weekly reviews together with the Connect team.

Here's how it works:

1. You load in the brief for a new assignment.
2. Sales Buddy pulls out the problem statement and the budget range.
3. The assistant searches the agency's own knowledge base for similar proposals and projects.
4. You validate the suggested references. The human decides, always.
5. From there you keep going with targeted questions, for example about the approach, the methodology or the phasing.

We built the assistant on Microsoft Copilot Studio, inside Connect's existing Microsoft 365 environment. That was a deliberate choice: the data stays where it already lives, the user works in an environment they know, and there's no separate tool needed. The knowledge base got a simple structure with labels per client, budget and category, so the assistant only pulls in relevant documents.

We started with a pilot group and wrapped up with documentation and a knowledge handover, so the Connect team can carry on independently.

The benefit: the team no longer starts from a blank page, but from the strongest past files.

Results

Sales Buddy is live today with a pilot group at Connect. What used to start with digging through folders now starts with loading in a brief and getting a list of relevant references. The stated goals are clear: roughly a quarter less time per proposal, a third less time across the full proposal process, and more reuse of strong strategies from past files.

Just as important is what happened alongside the tool. The path from mapping to business case to build gave Connect a shared language around A.I. The team has started building its own agents, and we keep supporting that as a partner.

Flowkify keeps supporting Connect as its A.I. partner, with a focus on scalable solutions and maximum impact.

Curious what A.I. agents could mean for your service organization? Get in touch with Flowkify for a tailored A.I. scan.