Writing a Public Tender Application with AI

The application is the permanent paradox of architecture firms. You present the same practice, the same partners, the same competencies, the same forty references -- and yet every application has to be rewritten. Not because your firm changed between Tuesday and Thursday, but because the tender regulations, the programme, the client, and the evaluation criteria are different every time.

The result: between 3 and 4 hours per application, a good chunk of which is spent reformulating what you've already written ten times. This is exactly the kind of task where AI changes the game -- provided you use it correctly.

This article is part of the complete guide on using AI for architecture tenders.

Why the application is the most repetitive part

Out of the 10 to 16 hours a full tender response takes, the application accounts for 3 to 4 hours. It's not the longest stage (the technical offer often exceeds it), but it's the most thankless. The reason is simple: 80% of the content is material you already have. Your firm presentation exists. Your references are documented. Your team is in place. The work isn't about creating information -- it's about reformulating, reorganising, and reframing it to fit today's brief.

And that's precisely what makes it painful. Reworking an existing text to adapt it is mentally harder than writing from scratch. You hesitate to rewrite everything, you do partial copy-paste, you adjust a sentence here, a paragraph there. The result is often a patchwork that lacks coherence.

AI doesn't tire of this work. It takes your inputs -- firm profile, references, brief criteria -- and produces a fresh text every time, structured and tailored to the context. This is its natural playing field.

The components of an application

Before we talk about AI, let's recap what a complete application contains. Depending on the tender, the package includes:

• The cover letter (or letter of application). The first document the evaluation panel reads, it must demonstrate that you understand the project, that your experience is relevant, and that your approach matches their expectations. It sets the tone and determines whether the rest of your submission will be read carefully or skimmed.

• The firm presentation. Headcount, turnover for the past three years, technical resources, certifications, partners' career summaries. This data changes little from one tender to the next, but its formatting must adapt to each brief's requirements.

• The references. A selection from your portfolio of the most relevant projects, presented with the expected metrics: floor area, construction cost, nature of the client, scope of services delivered, completion year.

• The team resources table. Composition of the proposed team, individual competencies, availability rates, tools and software used.

Each of these components has its own writing constraints. And each can be handled by AI in a different way.

Tailoring the cover letter to the specific project

The cover letter is the element that suffers most from copy-paste. You know it, you do it anyway. You take the one from the last tender, change the project name, adjust two sentences, and send it off. The panel can tell: a generic letter is spotted by the second paragraph.

AI lets you break out of this trap. By feeding it three elements -- your firm profile, the brief analysis, and the programme specifics -- it produces a letter that mirrors the client's expectations.

Concretely, if the brief emphasises experience in listed building renovation, the letter foregrounds your heritage references. If the client is a social housing developer, the vocabulary shifts: you're talking about cost control, tenant consultation, occupied-site management. If the programme highlights strong environmental ambitions, the letter articulates your bioclimatic approach.

This isn't a template with blanks to fill. It's a purposefully written text that reflects your actual positioning, oriented towards what the panel wants to read. You review, adjust the tone, add a personal touch -- but you start from a solid draft, not a blank page or a tired copy-paste.

Smart reference selection

Your firm has accumulated 30, 40, maybe 60 references over the years. The brief asks for 5, sometimes 3. This choice is strategic: the wrong references, even architecturally excellent ones, can drag your application down if they don't match the selection criteria.

This is an exercise where AI adds genuine value. By cross-referencing all your references against the criteria extracted from the brief -- programme type, construction cost, nature of the client, scope of services, location -- it can recommend the five most relevant references and explain its reasoning.

For example, for a tender to build a primary school in a rural area for a municipality of 3,000 residents, the AI will prioritise your public facilities of comparable scale, your projects with municipal clients, your rural experience -- rather than your striking residential tower in the city centre, even if it's your most high-profile project.

This selection doesn't replace your judgement. You know subtleties the AI can't perceive: a particular client you'll encounter on the jury, a reference that illustrates a skill that's hard to demonstrate any other way. But the AI's pre-selection saves you the time of systematic comparison and guards against blind spots -- that perfectly suited reference you'd forgotten at the bottom of your portfolio.

Structuring the presentation to mirror the criteria

Here's a habit few firms have, yet it makes a real difference: calibrating the weight of each section of your application to match the criteria weighting in the brief.

If the brief assigns 25% to team resources, 25% to references, 30% to methodology, and 20% to programme understanding, your application should reflect that breakdown. A three-line paragraph on team resources when the criterion carries a quarter of the score sends a signal to the panel: you either didn't read the brief carefully or have nothing to say on the subject.

AI applies this logic systematically. From the weighting grid extracted from the brief, it structures the firm presentation and calibrates the depth of each section. If team resources count for 25%, it develops individual competencies, career paths, availability, and specific experience for each proposed team member. If references count for 40%, it details each selected project with full metrics.

An experienced architect does this calibration intuitively -- but it takes time, and it tends to get neglected when you're under pressure. AI does it every time, without exception.

The commercial vocabulary trap

This is the most common and most damaging mistake when using a generic AI to draft an application. An unconfigured ChatGPT or Claude will naturally produce text that sounds "sales-y": "our innovative and passionate firm," "we are committed to delivering an exceptional client experience," "our creative and bespoke approach."

This register is completely off-target in public procurement. Technical evaluation panels for local authorities, social housing developers, and public institutions read dozens of applications. They spot marketing-speak instantly. And they draw a simple conclusion: this applicant doesn't understand the codes of public procurement.

The right register is factual, technical, and institutional. You write "the firm has proven experience in the construction of school complexes," not "our firm is passionate about school architecture." You write "the references below demonstrate competence in heritage building renovation," not "we have an impressive track record in heritage." You write "the proposed team holds the qualifications required by the specifications," not "our team of talents is ready to take on this challenge."

The difference may seem subtle on paper. When reading an application, it's immediately obvious.

How Super Archi calibrates the vocabulary

This is precisely the register problem that Super Archi solves. The Application assistant is configured with the specific vocabulary of architecture public procurement. It knows the acronyms and uses them appropriately. It distinguishes a client from a design team, a competition from an adapted procedure, a base scope from an extended scope.

More importantly, it produces text in the right register without you having to remind it with every prompt. The calibration is built in. You provide your inputs -- firm profile, references, brief analysis -- and the draft that comes out naturally uses the phrasing that a public procurement panel expects.

This isn't just about form. Accurate vocabulary signals to the panel that you're a professional who understands the framework they operate in. It's a credibility marker that works before the content is even evaluated.

The application is the front door. If the vocabulary rings false, the panel won't even read your references. An AI assistant configured for public procurement doesn't make that mistake.

The real time saving: from 3-4 hours to 1 hour

With a properly configured assistant, the application writing process drops from 3 to 4 hours to about 1 hour. Here's how that time breaks down:

• Preparing the inputs (10 min): you provide the AI with the brief analysis, your firm profile, and your reference list. If you use Super Archi, the firm profile is already loaded.

• Draft generation (5 min): the AI produces the cover letter, the structured presentation, the reasoned reference selection, and the team resources table.

• Review and adjustments (30-40 min): you review each component, correct nuances the AI missed, add your personal touch to the cover letter, validate the reference choices. This is the architect's work -- the part only you can do.

• Finalisation (5-10 min): final layout, compliance check against the brief requirements.

The gain isn't only quantitative. In one focused hour, you produce a more coherent submission than you'd have assembled in four hours of copy-paste and incremental adjustments. And crucially, the time recovered can be reinvested: responding to one more tender per month, or devoting more attention to the technical offer where the real differentiation happens.

To understand how this gain fits into the full tender response process, see the complete guide that covers every stage.

Ready to cut your application time by three? Super Archi includes a dedicated application assistant, calibrated for public procurement vocabulary and capable of producing a complete draft -- cover letter, references, firm presentation -- in minutes.

👉 Discover Super Archi