A good and effective robbery that isn't a typical adventurer's smash and grab has a number of components:
- Recon
- Infilitration
- Securing the Objective
- Exfiltration
These are all interesting and present exciting role-playing opportunities. Player may choose to preemptively plan or to assert that they have planned off-camera and make it up as they go along. The making it up as they go along must presume a plan off-camera, to explain the sudden appearance of resources that would have otherwise been available if they had planned it. This is, in many ways, recommended because so many players are so horrible at planning. Most players plans' presume that everything will go right and are entirely too "step 1, step 2, step 3" (to quote Pournelle). That kind of brittle plan will not function well in a robbery.
Instead, a resilient plan articulating objectives and resources and opposition will be far more effective in securing the objective. This sort of plan can be worked out after-the-fact, and supports a "I planned for this!" style of scenario (within limits). In a 4e sense, it may be worth running a skill challenge to plan which will give items and/or resources (from a diminishing supply) to rolls within the sneak. I would suggest against running a robbery as a skill challenge for the same reasons I would recommend against running an adventure as a skill challenge: the scope is wrong.
Recon is a matter of bluff, subterfuge, and misdirection. It is critical that the players learn that they cannot simply hit the place with "a bigger hammer" (Ringo) and have to use a sneak to accomplish their goals. The recon must also identify or allow the players to identify weak points. Instead of presenting players with a single route in and out, describe what the players learn about the various security systems in place (sentients count as a security system). While designing this, make sure to include some places where the defenses are weak and strong. Try to put yourself in the mindset of the designer and articulate what threats the system is set up to defend against. If it's strong against stealth approaches, it's probably weaker against a bluff approach. If the system is strong in all areas, be prepared for your players to give up.
Infiltration is a matter of successfully identifying weak points from recon and then systematically compromising them. Make sure that one failed roll will not effectively TPK the party, perhaps by having the alarm sounded via Big Red Buttons on the walls, and otherwise mostly isolating encounters and patrols. (Of course, if the players are stupid enough to leave bodies lying around then the "guys with guns" are called and it's game over.)
Securing the objective can be as easy or complex as you want it to be. Remember that no security system is perfect, and all locks are rated for time. If you want it to be more combatty, have the functional equivalent of a drill that needs to be defended from waves of patrols. If the point of the exercise is the infiltration and exfiltration, make the securing of the objective that much easier.
Exfiltration is, in many ways, the hardest part. A security system looks very different pointed the other direction, and at least some components of the system will be on alert due to flubbed rolls earlier. Good players will have alternate routes out of the security systems (teleporting is cheating, but can be an fun cheat so long as it takes effort to suborn the teleport-denial components) and this is also a fun time to have multiple countdowns going. The artifact could be a load bearing boss (warning, TVTropes), along with the various alarms being triggered and/or reset based on the various security breaches. Having everyone scrambling to escape and or get the artifact for themselves is a fascinating and tension-provoking exit, as is the "silent exfiltration" in the case of replacing the artifact with a good enough imitation and making sure that enough security systems remain quiet for the switch to not be noticed.
For systems to get inspiration from, take a look at:
- Leverage (It's about doing interesting robberies. It has plenty of stuff to steal for your game, especially the post-hoc plan feeling)
- Mouseguard: It presents ways to deal with and create complications. For a stealth game, the complications are the interesting component.