What we’ve built, what’s coming, and why BotMaster is becoming more than a platform
If 2024 was the year Dazerolab turned heads, 2025 is the year we started turning systems.
In eight months, we’ve gone from showcasing tailored automation projects to developing something bigger: a scalable, intelligent, and collaborative agentic workforce made of AI agents that can work together, learn from feedback, and adapt to real-world use cases.
We didn’t get here by chance. We got here by working backwards, as we always do: starting from pain points, not from blueprints. We listened, we tested, we rewrote. And now the results are starting to show.
More than AI: software that adapts, agents that learn
The shift from task-based automation to intelligence-based delegation isn’t just semantic; it’s structural. And with BotMaster, we’ve moved beyond AI as a feature: we’re shaping it into a network of operational agents that learn, evolve, and collaborate.
Each agent we deploy tells us more about how organisations think, how decisions are made, and where intelligence can actually make a difference. That’s why the story of 2025 is best told through the agents themselves.
Doc-Pro: from validator to decision-maker
Let’s start where most of our clients do: with Doc-Pro.
Initially developed to automate regulatory checks and document validation in the energy sector, Doc-Pro is is now proving its value across a broader range of use cases. Evolving through enterprise experience, it has become capable of handling everything from utility bills to legacy invoice formats and business customers with variable tax codes and subsidies.
It’s not just a tool. It’s an agent that evolves.
With every edge case it encounters, it becomes more precise. With every exception, it expands its logic. This isn’t just AI. This is operational adaptation.
Chat-Pro: the quiet powerhouse
Everyone imagines Chat-Pro as the interface: the friendly, smart assistant that powers companies’ conversations inbound and outbound. But behind the scenes, Chat-Pro is much more than a chatbot.
It’s a foundational agent for company knowledge. Stand-alone or integrated into MAS with other agents like Bid-Pro, it helps interpret queries (whether they are questions or tenders) using contextual data, domain-specific vocabulary, and structured inputs. It brings relevant content into the conversation and ensures every interaction aligns with how the company thinks and operates.
Its role is growing with every new deployment, especially as organisations start to see the value of conversational interfaces that are not only intelligent, but also deeply contextual.
The next wave: Bid-Pro and Pay-Pro
But what about the future months? They look no less than exciting!
First of all, we are about to launch Bid-Pro: a plug-and-play agent for parsing public tenders, managing RFP workflows, and creating competitive bid responses at speed.
While Doc-Pro is more direct and built to validate and execute with speed and precision, Bid-Pro is designed to articulate: it creates output that reflects strategy, not just structure. Each implementation of Bid-Pro is shaped as well around the company’s specific language, processes, and goals. No two bids are alike, and the agent evolves accordingly, supporting human decision-making while reducing the operational weight of the process.
To do so, Bid-Pro introduces a dual repository system:
- One for full document references;
- One for structured learning based on historical input/output pairs.
It also integrates Chat-Pro natively, creating a closed loop where every answer can be traced, refined, and redeployed. This is where MAS (multi-agent systems) stop being a theoretical model and start becoming a practical toolset.
On a parallel track, Pay-Pro is now in active development.
Its mission? To tackle one of the most painful areas in back-office operations: payment reconciliation. But that’s just the starting point. Pay-Pro is being built to support processes like installment management, overdue payment reminders, and even early-stage debt recovery, areas where timing, consistency, and clarity make all the difference. From validating payment evidence to flagging inconsistencies and managing delayed transactions, Pay-Pro aims to bring clarity and acceleration to one of the least enjoyable but most operationally critical parts of the business.
In its first iteration, Pay-Pro is being trained to handle a variety of formats and input sources, from structured APIs to loosely formatted attachments. The logic isn’t trivial. But that’s where we thrive.
The outlier: Claims-Pro
Not all agents come from roadmaps. Some emerge from experiments.
Claims-Pro was born during a hackathon: a travel-focused agent built to process large volumes of claims related to delays, lost luggage, and non-medical complaints. Built for speed and scale, it automates a process that typically drains time and resources.
More than a use case, Claims-Pro is a proof point: the agentic model is portable. It can be applied far beyond the original scope, in industries we hadn’t even targeted. And it still works.
From IT Day to Live sessions: shaping the voice of what’s next
2025 also marked a turning point in how we speak, not just what we build.
Our LinkedIn Live sessions have become a testing ground for new narratives, new terminology, new value propositions. Instead of pitching software, we explored pain points. We questioned assumptions. We showed our AI agents on the field while developing them, opened conversations and we met people who are now building with us.
At IT Day, we pushed things further. It wasn’t just another event; it was a turning point.
For the first time, we had the chance to step out from behind the screen and talk, in person, about what we’re building. We didn’t just present; we engaged. We answered tough questions, challenged assumptions, and saw first-hand what resonates, and what doesn’t.
There were conversations that turned into insights. Curiosity that turned into collaboration. Handshakes, questions, deep dives into use cases, all happening in real time, with real people.
For a company like ours, built around early-stage vision and deep tech execution, IT Day was a reminder: no matter how advanced the tech, it only matters when it meets real needs and the people behind them. And the best way to understand those needs is to be there, to listen, and to talk things through.
What to expect before the end of the year
The final stretch of 2025 won’t be just about stabilising what we’ve done. It’ll be about pushing the boundaries of what an agent can do and how fast it can be brought to life.
Over the next four months, we’ll bring Bid-Pro into production with selected partners, starting with organisations that have a clear pain point: bidding on tenders is slow, repetitive, and prone to inconsistency. Bid-Pro will not only automate, but also suggest, learn and refine, creating a virtuous cycle between legal, sales and operations.
At the same time, Pay-Pro will begin pilot testing in financial environments where timing and accuracy are paramount. We’re talking about real-time reconciliation logic, multi-source validation, and integration with legacy systems; the kind of work that doesn’t make headlines but saves entire departments from drowning in paperwork.
Claims-Pro is also paving the way for a new vertical within BotMaster: travel and logistics. With its ability to handle high volumes of non-medical claims quickly and consistently, it demonstrates how agentic processes can unlock value in sectors beyond our original scope.
Another major milestone is around the corner.
BotMaster can now be deployed directly into enterprise AWS environments, enabling faster integration, enhanced security, and full control for internal IT teams. And we’re going further: soon, clients will also be able to purchase agents and usage credits via AWS Marketplace, making the agentic model even more accessible without compromising on customisation or control.
Behind the scenes, new agent ideas are being shaped.
Clients are coming to us not with feature requests, but with questions:
“Can I automate this?”
“What if the agent handled just this slice of the workflow?”
“How fast can we test it?”
That’s how this new phase feels: less about product, more about potential.
Less about software, more about intelligence at work.
Got an idea? Let’s build it
We really do think that the agentic model works because it starts from problems, not products. That’s why we don’t offer prepackaged solutions. We co-design agents. We prototype fast. We adapt what we’ve already built to what your process needs.
If you have an idea, a challenge, or a use case stuck in a spreadsheet: let’s talk.
We’ll build the agent that gets it done. Together.