In short
- Peer learning means people learn from and with each other, without knowledge having to pass through a trainer first.
- Learning is not the same as training. Training is one way to organise learning. Learning is the mechanism that builds skills.
- It starts with safety. When you're allowed to be unfinished, you dare to ask. When you dare to ask, you dare to try.
- It's affordable, not free. The skills are already in-house, so there's no invoice. Time for learning still has to be protected.
- AI can be an assistant teacher. In research, it has spread the know-how of the best to others and made less experienced helpers better teachers. Used as a substitute for colleagues, it quietens the asking.
- Getting started takes little: four sentences, five session formats and a 30-day experiment. You'll find them all in this handbook.
Chapter 1
What is peer learning?
Peer learning, also called peer-to-peer learning, is learning in which people learn from and with each other. The teacher isn't an outside expert. It's a colleague who is a few steps ahead, or who is figuring it out at the same time.
At work, peer learning happens all the time. You ask the person next to you how to run a report. You show a new joiner how to handle a difficult customer. Three people try a new AI tool and tell the others what worked.
So the problem isn't a lack of peer learning. The problem is that it's invisible, random and unevenly shared. The person who dares to ask learns. The person who doesn't falls behind. And the person who knows doesn't know who needs what they know.
Learning is not training
Training is an event in a calendar. Learning is a change in what a person can do. An organisation can run lots of training and learn little. It can also learn a lot without a single training day.
| Traditional training | Peer learning | |
|---|---|---|
| Starts from | A training plan and a calendar | A learning need, a person's own wish |
| Timing | When the course runs | When the need arises |
| Teacher | An outside expert | A colleague who knows a bit more, or figures it out with you |
| Content | Generic, pre-packaged | Your own work, your own examples |
| Results show up | On the attendance list | In what people now do differently |
| Cost | An invoice plus the whole group's time | Working time, no invoice |
Training has its place. Peer learning doesn't replace deep expertise, mandatory training or certifications. It fills everything in between, and there's a lot of in-between in a working day.
Learning by doing
Skills grow by doing. Laurea University of Applied Sciences in Finland has built its whole pedagogy on a model it calls Learning by Developing: learning starts from a real development task from working life and is done together with partners. In a workplace this is even easier, because the real tasks are already on the table. Nobody has to invent them.
Chapter 2
Why now? AI changed the price of information
AI can explain almost anything. It can't try things for you.
Information is now cheap and available any time. Skill is not cheap. Skill is knowing what to do in your own work, with your own customers, today. It grows by trying, failing and hearing how someone else solved the same problem.
At the same time, skill needs are changing faster than any training calendar can keep up with. When a new tool goes live on Monday, nobody waits for the autumn training day. Someone on the team tries it on Tuesday. The only question is whether what they learned stays with them or spreads.
AI doesn't make peer learning unnecessary. It makes it more important. The more automation there is in work, the more valuable the things automation doesn't do: judgement, experience, trust and the ability to ask the right question.
AI also brings a new opportunity. As an assistant teacher, it helps people put into words what they want to learn, find the right person, and become better teachers themselves. Pedagogy is no longer a tool reserved for training professionals. The research evidence is already strong, and Chapter 9 walks through it.
Chapter 3
Four sentences learning can start from
The core of peer learning fits into four sentences. Each one is a step of its own.
This is what I'd love to learn.
The wish becomes visible. One sentence is enough, and an unfinished wish is fine.
Me too.
People interested in the same thing find each other. Demand builds up, and one person's wish becomes a shared one.
I could help.
Someone with the skill recognises themselves. Often they didn't know their skill was needed.
Shall we try it together?
A learning moment happens: a short session, a shared experiment, or two people doing the thing together once.
That's the whole model. You could start it tomorrow with sticky notes on the coffee-room wall. The hardest part is the first sentence, because it means admitting you don't know yet.
Why this works
- Learning starts from a need. When a wish is visible, the skill can find it. Nobody has to guess what others would like to learn.
- Roles change. The same person can be a learner in one thing and a guide in another.
- The person sharing doesn't have to know everything. Sometimes deep expertise is needed. Sometimes it's enough that someone has walked a few steps ahead. And you're allowed to say: “I don't know. Let's find out together.”
From wish to action
A wish isn't learning yet. Learning starts with the first action. Don't wait for the perfect course. Pick one small experiment, do it, and tell others what you learned. The path is short: wish → one experiment → shared lesson.
Chapter 4
Learning doesn't start with training. It starts with safety.
When people feel safe, they dare to think in drafts. When they dare to be unfinished, they dare to ask. When they dare to ask, they dare to experiment. And when they dare to experiment, space for learning is born.
- Safetyallowed to be unfinished
- Couragedare to ask
- Experimentationdare to try
- Learningskills grow
This isn't just a nice idea. Google's Project Aristotle studied 180 teams to find out what sets effective teams apart. Of the five factors it identified, the most important was psychological safety: the sense that you don't have to fear asking, making mistakes or being unfinished in front of others. Back in 1999, researcher Amy Edmondson had already shown that psychological safety is linked to how teams learn.
What safety means in practice
- Private by default, shared by choice. People decide who hears about their wish: just themselves, the team or everyone.
- No scoring, no rankings. When learning becomes material for evaluation, people start performing. A person under surveillance stops saying what they don't know yet.
- Managers go first. When a leader writes the first learning wish, everyone else gets permission.
- People know who they're learning with. A short introduction before a session lowers the threshold more than any tool.
Chapter 5
What peer learning saves, and what it costs
The most expensive training is the one nobody asked for.
- The skills are already in-house. Every team has someone who knows the spreadsheet formulas, someone who knows the customer's awkward system, someone who runs a good meeting. Sharing that doesn't come with an invoice.
- Learning hits the real need. When learning starts from a wish, nobody sits through a course that isn't relevant to them. The people in the room are the ones who wanted to be there.
- Learning happens at the right time. Things are learned when they're needed, not six months later when the need has passed.
- The teacher learns the most. In a study from Washington University in St. Louis (Nestojko et al. 2014), people who expected to teach material to others remembered and organised it better than those who expected a test. For the person sharing, teaching is also their own development.
- Knowledge doesn't walk out of the door. When skills spread across more people, one person's absence or departure doesn't stop the work.
What it costs
Peer learning isn't free. It's affordable. The difference matters. The price is paid in working time: the expert's preparation, the learners' sessions, and the time of someone who holds the threads together at the start. When Finnish health and social care professionals were asked about developing their digital skills, lack of time came out as the biggest barrier, and colleague support and protected time as the most important enablers (Kaitila & Koivisto 2024). If time isn't protected, the model withers quickly.
Chapter 6
How others do it
Peer learning isn't a new invention. Some of the world's best-known technology companies have built their learning on it.
Googler-to-Googler (g2g)
A volunteer network inside Google in which more than 6,000 employees teach their colleagues alongside their own jobs: courses, mentoring and learning materials. According to the Human Capital Institute, nearly 80 per cent of Google's internal training happens through the g2g community.
An organisation's biggest source of learning is its own people.
ShipIt
Atlassian
Once a quarter, employees put their day-to-day work aside for 24 hours and build something new. Teams form freely across departments. Alongside engineers you'll find designers, lawyers and finance people, from new hires to the leadership.
Permission to experiment, and to show what you made.
Peer learning swarms
Intunex / Skillhive
In Intunex's Skillhive work, one client organisation ran hundreds of peer learning processes a year over several years. People who wanted to learn came together, and someone with the skill shared it. Spark's Wish Wall carries the same idea forward, now supported by AI.
Whoever knows can teach. Everyone else learns together.
Google is Google. But g2g isn't built on budget. It's built on trust, permission and making skills visible. Those are within reach of every team.
Chapter 7
Five session formats you can start tomorrow
Peer learning needs a little structure, but not a heavy system. These five formats work both in person and online.
Wish wall
- Ongoing
- Whole team or community
- Starts everything else
- Everyone writes down one thing they'd like to learn. A wall, a channel or a team board will do.
- Others add “me too” or “I can help” to the wish.
- When a topic has someone who can help and a few learners, you agree a time.
Makes two usually hidden things visible: what people want to learn, and who could help.
15-minute demo
- 15 min
- Weekly
- 2–30 people
- One person spends 5 minutes showing what they did and what they learned.
- Others ask questions for 5 minutes.
- Last 5 minutes: what will I try myself this week?
Unfinished work is welcome. Regularity turns sharing lessons into a habit, not an event.
Troika Consulting
- 25–35 min
- 3 people
- Liberating Structures
- Everyone takes a minute to think of their own challenge.
- The person whose turn it is describes their challenge (1–2 min). The other two ask clarifying questions (1–2 min).
- The person with the challenge turns their back. The other two brainstorm solutions out loud (4–5 min).
- They turn back round and say what was most valuable (1–2 min). Then the next person takes a turn.
Everyone gets help and helps others. No outside expert needed.
1-2-4-All
- 12 min
- Any group
- Liberating Structures
- Think about the question alone (1 min).
- Continue in pairs (2 min).
- Two pairs together: what repeats, what differs (4 min).
- Each group of four shares one key idea with everyone (5 min).
Every voice is heard, including the quietest, and quickly, even in a large group.
After Action Review
- 15–30 min
- After a project or event
- AAR
- What was supposed to happen?
- What actually happened?
- Why was there a difference?
- What will we do differently next time?
Originally a US Army method. Every success and failure becomes a shared lesson, and nobody has to look for someone to blame.
Chapter 8
A 30-day experiment for your team
Don't plan a year-long programme. Try it for a month, then decide. This experiment suits a single team and needs no budget.
Make wishes visible
The manager starts with their own learning wish. Everyone can write one wish, but nobody has to. A wish can relate to work or to something completely new.
Connect wishes and skills
Look at which wishes got a “me too” and an “I can help”. Pick 1–3 topics and agree the first 30–60 minute learning sessions.
Do it and show it
Each group runs one experiment in its own work. At the end of the week, hold a 15-minute demo. Unfinished is fine.
Look back and decide
Run an After Action Review with the four questions. Decide what to keep, what to change and what to stop.
What to measure
- How many learning sessions actually happened. Attendance alone says nothing about learning.
- Did you learn something? Did it feel safe? Yes or no is enough.
- One thing the team now does differently.
Don't score individuals. The point of the experiment is to learn, not to evaluate people.
Chapter 9
AI as an assistant teacher
AI is a good sparring partner. It isn't a teacher or a supervisor. At its best, it's an assistant teacher that helps people learn from each other. Its job is to activate thinking, not to automate it away.
What the research says
Peer learning at work supported by AI has rarely been studied directly. But the closest studies from working life and tutoring show three roles in which AI strengthens learning between people.
- AI spreads the know-how of the best. When more than 5,000 customer support agents were given an AI assistant, productivity rose by 14 per cent on average and by 34 per cent among novices. According to the researchers, AI passed on the practices of the most skilled workers to others (Brynjolfsson et al. 2025).
- AI makes a helper a better teacher. In a randomised trial at Stanford, AI guided human tutors in real time. Students learned more, and the gains were largest among students whose tutors had been rated least effective (Wang et al. 2025). A teacher doesn't need to be a finished expert when AI supports them.
- AI connects different kinds of expertise. In a field experiment at Procter & Gamble, 776 professionals worked on real product development challenges. An individual working with AI performed as well as a two-person team without it, and AI broke down the silos between R&D and commercial staff (Dell'Acqua et al. 2025).
A warning: AI can also quieten the asking
After ChatGPT launched, activity on the programmers' Q&A site Stack Overflow fell by about a quarter within six months (del Rio-Chanona et al. 2024). People asked AI instead of each other. At work, the same thing means a colleague's know-how goes unused and new lessons don't spread. That's why AI should connect people, not replace them.
Where AI helps
- Putting the wish into words. “I'd like to know AI” becomes “I'd like to learn to draft customer replies with AI and check them myself.”
- Finding the right person. AI can notice that two people want to learn the same thing and a third knows how.
- Suggesting a first step. Where to start today, what to read, what to try.
- Supporting the teacher. AI can suggest an outline, examples and questions for a learning session. The organiser decides the content.
Where AI should not be used
- Scoring, classifying or profiling people.
- Deciding who should learn what.
- Replacing a colleague's experience. AI can also be wrong. People make the decisions, and meetings between people are the heart of learning.
Example
Spark's Wish Wall
Spark's Wish Wall is built on the idea behind this handbook. You write in one sentence what you'd like to learn. Spark turns your wish into a picture and suggests first steps and library tips. Others can join as learners, encourage you, ask questions or offer to help. When one person who can teach meets a few who want to learn, a learning session can happen.
Your wish is private until you choose to share it. Nobody is scored. A team board is visible only to invited members. Spark is still an early prototype, and we build it in the open. Read how safety is handled.
Chapter 10
How peer learning dies
You can kill peer learning with good intentions. These are the most common ways.
- It's made mandatory. Forced learning is just training on worse terms.
- People get scored and ranked. Then nobody admits what they don't know.
- Only experts are allowed to teach. Then there are too few teachers, and everyone else becomes an audience.
- No time is set aside. Learning gets crushed by busyness, and the experiment fades within weeks.
- You measure attendance, not learning. A full room doesn't tell you whether anything changed in the work.
- You plan a year-long programme before the first experiment. Start small and learn as you go.
The role of HR and L&D changes
Peer learning doesn't make HR or L&D unnecessary. It shifts the role from organising training to enabling learning. Three tasks are enough to start: make wishes and skills visible, protect time for learning, and remove barriers such as calendars, rooms and lack of permission. After that, the training budget goes where in-house skills aren't enough.
Frequently asked questions
What is peer learning?
Peer learning means people learn from and with each other, without knowledge having to pass through a trainer first. At work, the teacher is often a colleague who is a few steps ahead or is figuring it out at the same time.
Is peer learning the same as peer-to-peer or P2P learning?
Yes. Peer learning, peer-to-peer learning and P2P learning usually mean the same thing: people learning from and with each other. In workplaces the term often also covers peer coaching, mentoring between colleagues and knowledge sharing within teams.
What's the difference between peer learning and training?
Training is one way to organise learning. Learning is the mechanism that builds skills. Training usually starts from a plan and a calendar, peer learning from a learning need and real work. Peer learning doesn't replace deep expertise, mandatory training or certifications, but it fills everything in between.
Who can teach in peer learning?
Anyone who knows a little more than the others, or is willing to figure it out together. The person sharing doesn't need to know everything. Roles change: the same person can be a learner in one thing and a guide in another.
How much does peer learning cost?
Peer learning is affordable because the skills are already in-house and there is no invoice. It isn't free: the expert's working time, a regular rhythm and time for learning have to be protected. Lack of time is the most common barrier to learning at work.
Why does peer learning start with psychological safety?
People only learn if they dare to be unfinished: to ask, try and fail. In Google's Project Aristotle study of 180 teams, psychological safety was the most important trait of effective teams. The chain is: safety, courage, experimentation, learning.
Does peer learning work in remote and hybrid teams?
Yes, but it doesn't happen by itself, because corridor and coffee-break conversations are missing. Make learning wishes visible on a shared board and agree short, regular online sessions, such as a 15-minute demo once a week.
How do you measure the impact of peer learning?
Lightly and without scoring individuals: how many learning sessions actually happened, whether people felt they learned something and that it felt safe, and what the team now does differently. Attendance alone says nothing about learning.
How does AI relate to peer learning?
AI can act as an assistant teacher: it helps put a learning wish into words, find people who want the same thing, suggest a first step and outline a learning session. In research, AI has spread the practices of the most skilled workers to others and made less experienced tutors better teachers. AI should not score or profile people, because a person under surveillance stops saying what they don't know yet.
Will AI replace learning from colleagues?
No, and it shouldn't. AI can explain and suggest, but experience, judgement and trust grow between people. After ChatGPT launched, activity on the programmers' Q&A site Stack Overflow fell by about a quarter within six months, because people asked AI instead of each other. AI strengthens peer learning when it connects people.
Start with one sentence
Write down one thing you'd like to learn. See who says “me too”. For your team, you can set up a private board that only invited members can see.
Sources
- Human Capital Institute: Google's g2g “Googlers-to-Googlers” Program: A Lesson in Community, Culture and Trust.
- Atlassian: Atlassian's ShipIt hackathon for technical and non-technical teams.
- Google re:Work: Guide: Understand team effectiveness (Project Aristotle).
- Edmondson, A. (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly, 44(2), 350–383.
- Nestojko, J. F., Bui, D. C., Kornell, N. & Bjork, E. L. (2014). Expecting to teach enhances learning and organization of knowledge in free recall of text passages. Memory & Cognition, 42. Press release: Washington University in St. Louis.
- Kaitila, K., Koivisto, L. & Mukka, M. (2025). Vertaisoppimisen hyödyntäminen työyhteisön digitalisoitumisessa [Using peer learning in the digitalisation of work communities]. LAB Pro. In Finnish.
- Laurea University of Applied Sciences: Learning by Developing (LbD).
- Liberating Structures: Troika Consulting and 1-2-4-All. Timings: SessionLab, Troika Consulting.
- Brynjolfsson, E., Li, D. & Raymond, L. (2025). Generative AI at Work. The Quarterly Journal of Economics, 140(2), 889–942. NBER Working Paper 31161.
- Wang, R. E., Ribeiro, A. T., Robinson, C. D., Loeb, S. & Demszky, D. (2025). Tutor CoPilot: A Human-AI Approach for Scaling Real-Time Expertise. Stanford University.
- Dell'Acqua, F. et al. (2025). The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise. NBER Working Paper 33641.
- del Rio-Chanona, R. M., Laurentsyeva, N. & Wachs, J. (2024). Large language models reduce public knowledge sharing on online Q&A platforms. PNAS Nexus. Press release: Oxford INET.
- Sichterman, B. et al. (2025). Supporting peer learning with artificial intelligence: a systematic literature review. Innovations in Education and Teaching International, 62(5).
- Spark's Wish Wall: About, Security, For teams, Help & FAQ.