Coming together for our Autumn company update

Deep sector expertise - Myself and Lisa

Lisa and I opened the day with an update on our refreshed sector-led approach, following our restructure earlier this year. This new approach allows us to focus more deeply on the distinct challenges that the charity and public sector face, and to make sure the work we’re doing has the greatest impact for our clients.

This shift has helped us build stronger relationships and have more meaningful conversations. In the public sector, we’ve built real momentum through events like the Global Government Forum and the Local Government Technology Conference, where we shared insights from our work with Guy’s and St Thomas’ NHS Foundation Trust, and gave demos of our AI-powered qualitative research tool, Sonar, which we also hosted as a webinar.

Lisa and I also shared insights into our work with clients like Tate and London Museum, who are both using AI and open source technology to connect more deeply with their audiences.

In the charity sector, we’ve hosted webinars and Breakfast Events on topics such as how AI is changing search for health charities and how to engage Gen Z supporters. These sessions bring together charity leaders to share ideas and challenges, helping us all learn from one another and stay close to what’s happening in the sector.

All of this is only possible thanks to the brilliant people across Torchbox who share their expertise through blogs, talks, events, and everyday collaboration. Their insight helps us show up in the right spaces, stay connected to the sectors and causes we care about, and continue to grow.

Looking ahead, we’ll keep building on this: working with clients to tell their stories, creating genuinely useful content, and contributing to the conversations that are shaping the future of digital in the charity and public sectors.

Future-ready services - Nick, Dan and Olly

Next, Nick, Dan and Olly shared updates on how we’re developing services for the future, including three areas where we’re already experimenting and learning.

Caddy (Nick)

Nick introduced Caddy, a project that emerged from the GOV.UK AI programme, and builds on tools we’ve already been experimenting with, like Langchain, LiteLLM, and AWS services. The idea is to develop modular, reusable components that can support government teams exploring AI. Some of this work could eventually be open-sourced, creating opportunities for Torchbox to contribute expertise and help shape responsible approaches to AI in the public sector.

Data Brain (Dan)

Dan shared an update on DataBrain, one of our Innovation Fund projects. It tackles a familiar challenge: reporting campaign performance when data is scattered across multiple platforms. By bringing data into one place, it frees up time that would otherwise be spent on manual exports and gives teams more space to focus on insights and action.

AI services (Olly)

Olly spoke about the growing number of clients looking for support with AI. Some are already experimenting with tools, while others are cautious but curious. He shared more about how we’re developing consultancy and product offers to meet organisations where they are, helping them take practical, safe steps that add value without overpromising.

Wagtail AI - Thibaud and Sage

Thibaud and Sage shared the latest progress on Wagtail AI, guided by clear principles: AI should be opt-in, responsibly adopted, model and provider-agnostic, and with a human in the loop.

They demoed three new features in development:

• Page title and SEO descriptions - suggesting titles and meta descriptions based on page content, which editors can then edit or regenerate.
• Image descriptions - generating alt text at upload and in-context, helping address the huge backlog of images without descriptions.
• Content feedback - giving inline suggestions for clarity, tone or structure, while keeping the final decision with the editor.

A lot of thought is going into accessibility, bias, and environmental impact, for example, choosing lower-footprint models wherever possible. Beyond these features, the team is also working on reusable UI components for Wagtail core, a Django AI Core package for the wider Django community, and a standalone Wagtail search module.

Revolutionising our processes - Jess

Jess shared a case study from our work with a recent charity client, where the delivery team experimented with new ways of working.

Some of the things they tried included:

• A small, consistent core team with one main client stakeholder.
• A shared Slack channel with the client to make all communication transparent.
• Using Fathom AI notetaker to be more present in the meetings.
• A “good enough” mindset for tickets and planning – just enough detail, just in time.
• Encouraging the client to use the platform daily, instead of relying on formal show-and-tells.
• Open-source UI kits and reusable Django components for faster, accessible builds.
• AI to speed up engineering, always with human oversight.

Jess was also candid about what she’d do differently next time: setting clearer rules of engagement for experiments, pushing for more stable team setups, and striking a better balance between lean tickets and forecasting.

James and Holly - AI Accelerator

Holly shared an update on our AI Accelerator, a company-wide programme focused on improving how we work day to day. The goal is to explore where AI and automation can help us save time, improve quality, and make space for higher-value work, while always being responsible and ethical.

Earlier this year, we ran a company survey to understand how confident people felt about using AI in their roles. We revisited that same survey during the session, and it was great to see how much things have moved on. Confidence scores have risen across the board, showing that while we’re still learning, more people are starting to use AI tools in ways that genuinely help them in their work.

James then shared a few real examples of that progress.

• Our Statement of Work Generator aims to save 25 hours a month by automating parts of the scoping process.
• A new credentials tool is helping us quickly find case studies and project details across torchbox.com, making it easier to pull together examples for proposals and pitches.
• And our GTM Audit Tool turns complex tag manager data into clear reports in minutes, improving accuracy and saving time.

Together, these experiments are helping us build the foundations for more efficient, informed ways of working.

Looking ahead, the focus will be on helping teams embed these tools into their everyday processes, identifying new opportunities, and continuing to build confidence through training and collaboration.

Our trustees & the Torchbox Constitution - Helen, Johnny, Lisa, Olly

Finally, we heard from our trustees. Helen, Johnny, Lisa, and Olly introduced the new Torchbox Constitution, which sets out our purpose, principles, and governance as an employee-owned company. It’s designed to make rights and responsibilities clear, and to act as a guiding reference when questions arise about how we work or make decisions.

They’ll share more about this in a follow-up blog soon.

Wrapping up

Days like this are a good reminder of what we’re building together. We’re clearer on our sector focus, making progress on AI in a thoughtful way, and finding better ways to deliver our work, while keeping hold of the things that make Torchbox, Torchbox.

If any of this sparked ideas or questions, I’d love to hear them. You can find me on LinkedIn or get in touch via the form below.